Design of a Humanoid Biped for Walking Research

by

Daniel Joseph Paluska

Submitted to the Department of Mechanical Engineering in partial fulllment of the requirements for the degree of Master of Science at the MASSACHUSETTS INSTITUTE OF TECHNOLOGY September 2000 c Massachusetts Institute of Technology 2000

Signature of Author . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Department of Mechanical Engineering August 31, 2000 Certied by . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Gill A. Pratt Assistant Professor of Electrical Engineering and Computer Science, MIT Thesis Supervisor Certied by . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ernesto Blanco Adjunct Professor of Mechanical Engineering, MIT Thesis Supervisor Accepted by . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ain Sonin Chairman, Departmental Committee on Graduate Students

Design of a Humanoid Biped for Walking Research by Daniel Joseph Paluska

Submitted to the Department of Mechanical Engineering on August 31, 2000, in partial fulllment of the requirements for the degree of Master of Science

Abstract
This thesis presents the design of the robot M2. The primary motivation behind the work is to create a platform for research into bipedal walking. Bipedal robots have been built in the past, but most of them relied heavily on lessons from robotic arm research. There is evidence that suggests actuation and design for legged locomotion should deviate from the past standard of robotic arm design. Specically, there should be an emphasis on low impedance actuation, shock tolerance, passive control mechanisms, and weight reduction. Human data which supports these points is presented. The robot M2 was designed with these things in mind. M2 is a humanoid bipedal robot with twelve active degrees of freedom. It is essentially a torso and two legs and with linear dimensions of a 50th percentile US male. It weighs approximately 60 lbs(25kg). All the active degrees of freedom are powered using Series Elastic Actuators, which provide force control and shock tolerance. For simplicity, degrees of freedom above the hip are absent. Thesis Supervisor: Gill A. Pratt Title: Assistant Professor of Electrical Engineering and Computer Science, MIT Thesis Supervisor: Ernesto Blanco Title: Adjunct Professor of Mechanical Engineering, MIT

Acknowledgments
The development of the robot has been a three person project from the beginning. Jerry Pratt and David Robinson worked on the robot with me. Jerry performed preliminary simulations to determine feasibility of mass distributions and power requirements. David Robinson is responsible for the actuator design. I also had lots of help from Ben Krupp and Chris Morse who each had there own robot to build in the process. Pete Dilworth contributed plenty of tips from his own bipedal robot building experience. Allen Parseghian has been working hard with me trying to get the robot to actually walk. John Hurst has helped with debugging some actuator problems and designing a battery rig. Thanks to Mike Wessler and Andreas Hofmann for creation of and help with software tools. Thanks to Chris Barnhart for developing the DSP platform and handling my ignorance of all things DSP. During the construction, I had help from many people. We had several soldering parties in the lab which involved people sitting around the main lab table soldering. Jerry, Ben, Chris, Mike, Gaddy, Pete, and others participated in these. In addition, Jess helped me with cable making and was an immense help with my supposed native language. Professor Pratt has been a wonderful advisor. I am constantly amazed at how he has always placed his students rst. I have loved working in the lab and getting to play around! Special thanks goes to my friends and family who have made the time outside the lab great! Youve kept me sane during the worst of thesis despair.

This research was supportd by DARPA under contract number N39998-00-C-0656.

Contents
1 Introduction 1.1 Goals of Thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 Summary of Thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Human Characteristics 2.1 The Human Form . . . . . . . . . . . . . . 2.1.1 Weight Distribution . . . . . . . . . 2.1.2 Link Lengths . . . . . . . . . . . . . 2.2 Human Motion . . . . . . . . . . . . . . . . 2.2.1 Energetics of Human Walking . . . . 2.2.2 Human Kinetic and Kinematic Data 3 Robots and Actuators 3.1 Robotic Arms vs. Robotic 3.2 Passive Dynamic Walkers 3.3 Powered Bipedal Walkers 3.4 Series Elastic Actuators . Walkers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 9 10 11 11 11 11 14 14 15 16 16 16 17 17 19 19 19 19 22 22 22 24 25 27 27 27 27 29 29 29 31 31 31 31

4 Robot Mechanical Design 4.1 General Robot Architecture 4.2 The Design . . . . . . . . . 4.2.1 Overall Structure . . 4.2.2 Actuators . . . . . . 4.2.3 Ankle . . . . . . . . 4.2.4 Foot . . . . . . . . . 4.2.5 Hip . . . . . . . . . 4.2.6 Knee . . . . . . . . .

5 Electronics and Control Systems 5.1 Electronics Overview . . . . . . . . . . . . 5.1.1 48 Volt bus . . . . . . . . . . . . . 5.1.2 Robot Sensors . . . . . . . . . . . 5.2 Custom Circuit Boards . . . . . . . . . . . 5.2.1 The Signal Conditioning Boards . 5.2.2 The Analog Force Control Boards 5.2.3 The Analog Breakout Board . . . 5.2.4 The Power Distribution Boards . . 5.2.5 The DSP and Analog I/O . . . . . 5.3 Control Software . . . . . . . . . . . . . .

CONTENTS

A Joint Math A.1 Knee Joint . . . . . . . . A.2 Ankle Joint . . . . . . . . A.3 Hip Joint . . . . . . . . . A.3.1 Hip Pitch . . . . . A.3.2 Hip Roll and Yaw B Electrical Schematics

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

34 34 36 37 37 37 39 50 53

C Suppliers and Costs C.1 Costs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

List of Figures
2-1 The X velocities of the various parts of the body during a normal walking cycle. Adapted from data in Winter(20). . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-2 The frontal plane dimensions of a 50th percentile U.S. male. Adapted from data in Dreyfus(2). All dimensions are in inches. . . . . . . . . . . . . . . . . . . . . . . . . . 2-3 Several strides of the compass gait shown with corresponding potential and kinetic energy curves. From Krupp(8). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3-1 Some previous powered bipedal walking robots. . . . . . . . . . . . . . . . . . . . . . 3-2 An annotated rendering of the Series Elastic Actuator which is used in the robot M2. Modeled and rendered by David Robinson. . . . . . . . . . . . . . . . . . . . . . . . 4-1 A photo of the completed robot and a joint schematic view of the robot. The schematic shows active degrees of freedom only. The optional passive toe joint is not shown. . . 4-2 The dimensions of the biped in the frontal plane. The body and links are approximately axially symmetric about their longitudinal axis. Dimensions are in inches. . 4-3 A schematic of the ankle joint actuation scheme. The axes shown are xed to the center of the universal joint. The points A,E, and O are referenced in Appendix A. . 4-4 The dimensions of the biped foot with passive toe joint. The Xs represent the location of the load cells. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-5 A schematic of the biped hip. The pitch actuator is not shown. It lies along the axis of the thigh. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4-6 A photo of the biped thigh illustrating the hip pitch cable drive system. . . . . . . . 4-7 The ranges of motion on the robot ankle joint. The ankle roll is symmetric(+- 20 deg). 4-8 The range of motion of the robot knee. The knee has 80 degrees of motion. . . . . . 5-1 An overview of the biped electronics. Thick lines indicate power transfer and arrowheads indicate information ow. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-2 A photo of M2 with annotations of electrical systems. . . . . . . . . . . . . . . . . . 5-3 A photo of the biped strain gauge signal conditioning board. This board handles two strain gauges. There are a total of four of these boards on the robot. . . . . . . . . 5-4 A photo of the analog force control and joint pot buer board. There are six identical boards on the robot. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A-1 The robot knee joint with superimposed lines and points. Points O, A and M all refer to pin joints. O is the knee joint. M is where the actuator is mounted to the thigh and A is where the actuator is attached to the shin. . . . . . . . . . . . . . . . . . . A-2 A line drawing of the knee for the calculation of knee actuator desired force. The actuator is the line segment M A. This drawing also pertains to the geometry of the hip and ankle joints but the hip and ankle actuators have motions out of the plane whereas the knee actuator is always in the plane perpendicular to the knee axis. . . 12 13 14 17 18 20 21 23 23 24 24 25 26 28 29 30 30

35

35

LIST OF FIGURES

A-3 A simple view of the actuation of the robot ankle. This is a tranverse plane slice of the robot ankle. Point O is the intersection of the X and Y axes as well as the center of the ankle universal joint. Points A1 and A2 represent the attachment points of the two actuators respectively. Forces in the actuators create torques about both the X and Y axes. Joint rotations change the relative lengths of the X and Y projections(rx and ry ) of the moment arms, therefore aecting the torque to force transformation. . A-4 The robot hip joint schematic. Point O refers to the hip universal joint. The Z(hip yaw) and Y(hip roll) axes intersect at point O. Point Mr refers to the universal joint on the body where the roll actuator is mounted. Point Ar refers to the ball joint on the thigh where the roll actuator is attached. Point My refers to the pin joint where the yaw actuator is attached to body. Point Ay is the pin joint attachment of the yaw actuator to the yaw universal block. It is hidden in this schematic. . . . . . . . . . B-1 B-2 B-3 B-4 B-5 B-6 B-7 B-8 B-9 Page 1 of 6 of the biped force board. . . . . . . . . . . . . . . . . . . . . . . . . . . . Page 2 of 6 the biped force board. . . . . . . . . . . . . . . . . . . . . . . . . . . . . Page 3 of 6 of the biped force board. . . . . . . . . . . . . . . . . . . . . . . . . . . . Page 4 of 6 the biped force board. . . . . . . . . . . . . . . . . . . . . . . . . . . . . Page 5 of 6 of the biped force board. . . . . . . . . . . . . . . . . . . . . . . . . . . . Page 6 of 6 the biped force board. . . . . . . . . . . . . . . . . . . . . . . . . . . . . Page 1 of 2 of the strain gauge conditioning board. . . . . . . . . . . . . . . . . . . . Page 2 of 2 of the strain gauge conditioning board. . . . . . . . . . . . . . . . . . . . Page 1 of 1 of the analog I/O breakout board. This board is designed to interface with the ANA070 and ANA064 from Digital Designs and Systems. . . . . . . . . . . B-10 Page 1 of 1 of the M2 power distribution board. This board has two dierent population options. One to power the DSP and one to power the Intersense IS-300 tracker.

36

37 40 41 42 43 44 45 46 47 48 49

List of Tables
2.1 Human Mass Distribution. From Dempster(3). . . . . . . . . . . . . . . . . . . . . . 2.2 Human walking parameters from normalized data contained in Human Walking(18). Forces and power calculated for a for 1.83 m(62), 80 kg(178lb) person. Table displays non-concurrent maximum values which occur during an average walking cycle. . . . 4.1 Robot Joint Specications. Torque and rad/s numbers are given for maximum moment arm. Power, torque, and velocity are symmetric due to the actuator. *The ankle roll and pitch are not independent. Their maximum values can not be applied simultaneously. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Approximate distribution of mass in humans, the biped M2, and McGeers kneed passive dynamic walker(PDW). Human data adapted from Dempster and Gaughran (3). Robot weight distribution is driven primarily by actuator locations. Each actuator is approximately 1.2kg(2.5 lbs). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1 5.2 Circuit Boards on the Robot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sensors on the Robot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

15

19

22 27 28

C.1 Basic robot budget. See Robinson et al.((17)) for more detail on the actuator. This budget does not include prototyping or development costs. Part quantities and individual costs are not necessarily meant to imply identical parts but rather to give an average price for all units of a certain type. . . . . . . . . . . . . . . . . . . . . . . .

54

Chapter 1

Introduction
1.1 Goals of Thesis

The goal of this research project was to design a three dimensional, free standing bipedal robot. The main goal of the biped, M2, is to be a testbed for control ideas and walking research. Originally the robot was not specically humanoid, but after a bit of research we determined it would be the best course of action for several reasons. There is a vast amount of data regarding human form and motion. The robot can be considered a human replacement for hazardous situations. The robot will be large enough to avoid problems of miniaturization. Ideas regarding human control of locomotion can be tested. This thesis concentrates on documenting the mechanical portion of the project. The design of the robot was guided by several specic goals which include walking 1 m/s, climbing normal stairs, looking biological, turning dynamically, a three year life-span, and ten hours working time between mechanical or electrical failures. The following points dominated the design decisions of the robot. 1. Series Elastic Actuators Series Elastic Actuators(12) are used for all of the active degrees of freedom. These actuators provide force control as well as shock tolerance. Evidence suggests both factors are necessary for the task of biologically similar walking. The low output impedance of the actuators allows us to take advantage of the robots natural dynamics. All joints employ the same actuator design to minimize complexity and facilitate repairs. 2. Human Proportions The use of human proportions allows for easy comparison with biomechanics data. Human sizing also allows for use of large, standard components which are easy to see and debug. Also, by using human proportions, the research stays focused on walking and not on miniaturization. 3. Lightweight The robot frame is carbon ber and most other components are plastic or aluminum. The necessary actuator forces are kept low. A less massive robot is more manageable in a research environment. It is easier to handle and less likely to damage itself or harm researchers. 4. Mechanical Control Mechanisms Each joint (most importantly the knee) has adjustable stops with rubber pads. The foot of the robot is equipped with a passive toe joint. This joint has an adjustable range as well as a return spring. The limit stops are essentially high frequency non-linear PD loops which are dicult to implement in digital control even with the use of sophisticated electronics and sensors. The low impedance actuators allow for uncertain 9

CHAPTER 1. INTRODUCTION

10

contact, an inevitable occurence during walking on rough terrain. These mechanical features eliminate the need to operate any digital high frequency control loops on the robot. The robot is a three dimensional continuation of the work that began with the planar robot Spring Turkey(13) and continued with the planar robot Spring Flamingo(15). The previously enumerated points are the key areas where M2 diers from some other current three dimensional humanoid walkers(6, 21) which rely more on high impedance actuation and trajectory following control schemes.

1.2

Summary of Thesis

This thesis presents the design of the humanoid robot M2. Chapter 2 describes the human mechanisms and the forces, torques, velocities, etc. which are exhibited during a normal walking cycle. The human is a good starting point for the design of the robot. Human locomotion is well documented and there is a wealth of information on the human walking motion. Chapter 3 describes previous robots and actuators. It presents some of the dierences between robotic arm and walking robot technology. There is a brief introduction to Series Elastic Actuators. Chapter 4 describes the mechanical specics of the robot. Chapter 5 details the electronic, sensor, computer, and control systems on the robot. Chapter 6 gives some conclusions and some advice for future research. Appendix A describes the transformations from joint torques to actuator linear forces. Appendix B contains all the schematics for the custom circuit boards on the robot. Appendix C has a list of all of the suppliers that were used.

Chapter 2

Human Characteristics
There is a large database of knowledge regarding both the human form and human motion. We believe that mechanical characteristics found in humans aid in the control of walking and we should try to consider which of these characteristics can be embodied with current mechanical and robotic technologies.

2.1
2.1.1

The Human Form

Weight Distribution

An approximation of the human weight distribution is shown in Table 2.1. From this we can see that nearly 70% of the mass in humans is located at the waist and above. Figure 2-1 shows the linear velocities(in the walking direction, X) of dierent body areas. It can be seens that the body has the least uctuation in forward velocity and the foot has the most uctuation. Since kinetic energy is a direct function of velocity, the lower extremities have the greatest uctuations in kinetic energy. It is clear that more mass at the feet means more energy change. If this energy is all lost it would be very costly. In ecient walking, the mass is able to resonate with gravitational or elastic potential energy. An example of this can be seen in the compass gait described in the next section.

2.1.2

Link Lengths

The dimensions of a 50th percentile male are shown in Figure 2-2. The center of gravity is just above the hip at approximately 38.

Table 2.1: Human Mass Distribution. From Dempster(3). Body Area Head, neck, Shoulders, Thorax Arm Ab/Pelvic Thigh Shin and foot Percentage per Part 31% 5% 27% 10% 6% No. of Instances 1 2 1 2 2 Total Mass Percentage 31% 10% 27% 20% 12%

11

CHAPTER 2. HUMAN CHARACTERISTICS

12

ribcage hip shin top knee ankle toe

Velocity(m/sec)

0 1 5 10 15 20 25 30 35 40 45 50 55 60 65 70

-1 Frame Number

Figure 2-1: The X velocities of the various parts of the body during a normal walking cycle. Adapted from data in Winter(20).

CHAPTER 2. HUMAN CHARACTERISTICS

13

69.10 6.10 60.20 56.70

7.00 (Hip Spacing) 14.20 36.50

16.70

19.80

16.60

3.20 3.90 0

Figure 2-2: The frontal plane dimensions of a 50th percentile U.S. male. Adapted from data in Dreyfus(2). All dimensions are in inches.

CHAPTER 2. HUMAN CHARACTERISTICS

14

KE PE

Figure 2-3: Several strides of the compass gait shown with corresponding potential and kinetic energy curves. From Krupp(8).

2.2
2.2.1

Human Motion
Energetics of Human Walking

Data in Rose et al.(18) shows us that a 180lb person walking at a normal pace consumes about 320 Watts of power. It is estimated that roughly one quarter of that is mechanical power. This means that at a comfortable pace, a 180lb human can walk with only 80 Watts of mechanical power. This can be a somewhat depressing fact considering one of the electric motors used on the robot M2 can output 90 Watts continuously - and there are 12 motors on the robot! Hopefully research on M2 will allow future robots to walk as mechanically ecient as a human. A simplied human gait known as a compass gait can give us insight into the energetics of walking. The compass gait model consists of two rigid, massless legs attached at the hip with a pin joint. A point mass is modeled at the hip joint. With the only degree of freedom located at the hip, the body is forced to follow an arc dened by the length of the leg. The compass gait can be seen in Figure 2-3. The compass biped has instantaneous, discretely changing dynamics. It simply changes from one inverted pendulum to another when the swing leg touches the ground. Calculating the external work on the model, namely the potential and kinetic energy of the system PE KE = M gLcos = M gL(1 cos) (2.1) (2.2)

By graphing the equations 2.1 and 2.2 in Figure 2-3, the cyclical nature of potential energy and kinetic energy can be seen. Note that the kinetic energy is maximum when the potential energy is minimum and vice versa. Similar cyclical transition of potential and kinetic energy was recorded

CHAPTER 2. HUMAN CHARACTERISTICS deg 30,-18 8,-7 5,-15 68,8 10,-15 na rad/s 3.6,-2.0 1.6,-1.0 4.0,-3.0 5.8,-7.8 3.0,-4.2 NA Nm -111.0 -63.5 8.0 -71.4 -63.5 40.0 W 56 +-28 -16 -79.5 280 -16

15

hip(pitch) hip(roll) hip(yaw) knee(pitch) ankle(pitch) ankle(roll)

Table 2.2: Human walking parameters from normalized data contained in Human Walking(18). Forces and power calculated for a for 1.83 m(62), 80 kg(178lb) person. Table displays nonconcurrent maximum values which occur during an average walking cycle. in human subjects by McMahon (11) using force plate data to calculate the changes in mechanical energy of the bodys center of mass. The compass gait is similar to the behavior found in the passive dynamic walkers described in the next section. There is also evidence that humans use some passive dynamics during the gait cycle. At the beginning of swing phase, the leg receives a tiny impulse of energy and then the natural pendulum motion carries it forward to extension. The leg receives another tiny impulse of energy to bring it to a sudden halt before touchdown. Electromyographic data record by McMahon (11) shows little electrical activity in the leg muscles of humans during the swing phase at normal walking speeds. This suggests that the leg is swinging freely during this period. In addition, electromyographic records show signicant electrical activity in leg muscles during stance.

2.2.2

Human Kinetic and Kinematic Data

There is a wealth of data from recordings of human motion. These recordings have been transferred into data regarding joint motions and joint torques during the walking cycle. In addition, the author relied on a rather low-tech method for extracting velocity data from the graphical position data. Power, torque, position, and velocity data for an average human walking cycle are shown in Table 2.2.

Chapter 3

Robots and Actuators

3.1 Robotic Arms vs. Robotic Walkers

Many high performance robotic arms and hands have been developed for use in factories, space, and research(19). It might seem to an outside observer that these technologies could be exploited for use in a legged robot. Most of the time this is not the case. There are several reasons why. Fixed vs. Floating Reference Robotic arms are generally xed to an inertial reference frame(factory) or a body whose mass is large enough that it can be considered xed(spacecraft). A walking robot is not xed to any reference frame and has a limited set of torques which it can apply due to its limited contact with the world. Onboard vs. Oboard Robot arms can often place their heavy motors at their xed end. Then the motors are only responsible for moving the frame of the arm and not themselves. Because a walking robot must carry all its components, the motors support themselves as well as the structure of the robot. Carrying power is also an issue for walking robots although most are tethered due to battery limitations. Environmental Awareness Robot arms are not usually expected to perform in unknown situations. They generally are designed for specic working conditions and their ability to handle unexpected disturbances is limited. Ideally, walking robots are supposed to handle rough and unknown terrain. Success Metrics Robotic arms are often judged on their ability to position their end eectors precisely. Robotic walkers are usually not judged on their ability to position precisely but rather on their ability to get from Point A to Point B without falling down. Impacts Most robot arms are not designed to handle impacts. Walking, however, has an impact at every touchdown.

3.2

Passive Dynamic Walkers

Much research has been done on completely passive walking machines (5, 1, 4). Passive dynamic walkers use Earths gravity as a power supply. They rely on special geometry and mechanical mechanisms as control systems to achieve gaits very similar to the compass gait depicted in Figure 2-3. Though they are not robots in the traditional sense, they can give insight into walking machines. These robots exploit passive dynamic elements and mechanical mechanisms to achieve highly ecient gaits. In the limit, a passive dynamic walker relies entirely on passive dynamic motions and thus requires no external power source other than a small incline. Several passive legged robots such as these have been constructed and succesfully demonstrated (5, 1, 4, 9).

16

CHAPTER 3. ROBOTS AND ACTUATORS

17

Figure 3-1: Some previous powered bipedal walking robots. From top left to bottom right are WL-10RV1 from Waseda, P2 from Honda, Toddler from UNH, the Moscow State University Biped, SD-2 from Clemson and Ohio State, Biper from University of Tokyo, Meltran II from Mechanical Engineering Lab in Tsukuba, and Timmy from Harvard. Figures compiled by J. Pratt.

3.3

Powered Bipedal Walkers

Many bipedal walking robots have been built over the years. Several of these robots are shown in Figure 3-1. These biped walking robots fall into two broad categories: those which predominantly use pre-recorded trajectory playback and those which predominantly use an algorithmic controller. All the robots employ a high impedance actuation scheme and none take advantage of natural dynamics. Passive dynamics were successfully implemented in actuated robots as well. J. Pratt (16) developed algorithms that used the natural dynamics of the swing leg to eciently control a planar bipedal robot. More recently, J. Pratt has developed three-dimensional simulations of M2 using passive elements such as a knee cap, a compliant ankle, and a passive swing leg to achieve natural looking and ecient walking.

3.4

Series Elastic Actuators

Series Elastic Actuators(SEAs)(14) are actuators which have an elastic element in series with the motor and gear train. A sensor measures the displacement of the elastic element and force is implied by Hookes Law, F = kx. In short, SEAs provide force control, shock tolerance, and low impedance actuation. A more detailed account of their benets and limits can be found in Robinson et al.(17) A rendering of the SEA used on M2 can be seen in Figure 3-2. The brushless DC motor is modied to have a ballscrew as its rotor. The output of the ballscrew is attached to an aluminum

CHAPTER 3. ROBOTS AND ACTUATORS

18

brushless DC motor

ballscrew

carbon fiber tubes linear bushings linear potentiometer

springs

output shaft

machined aluminum pieces

Figure 3-2: An annotated rendering of the Series Elastic Actuator which is used in the robot M2. Modeled and rendered by David Robinson. piece which is sandwiched between a set of linear compression springs. The other ends of the springs are attached to the output of the actuator. A linear potentiometer measures the ballscrew displacement with respect to the output shaft, thereby measuring the spring compression and giving an indication of the force output. On the specic SEA designed for M2(17), a simple analog PD controller is implemented to control the spring deection. When the output of the actuator is clamped, a force control bandwidth of 30hz is observed. The actuator is capable of outputting 300 lbs and capable of resolving forces on the order of 1.5 lbs.

Chapter 4

Robot Mechanical Design

This chapter presents the design specics of the bipedal robot M2.

4.1

General Robot Architecture

The robot stands approximately 5 feet high and weighs 60 pounds. The robot has 12 active degrees of freedom. Table 4.1 lists the specications for the joints of the robot. Power at each joint is the same due to the single actuator design employed for all the joints. The power to the individual degrees of freedom of the ankle joint is limited by the power of the other. The robot joint ranges of motion were chosen based on data from humans as well as the robot Spring Flamingo and preliminary dynamic computer simulations.

4.2
4.2.1

The Design
Overall Structure

A photo of the robot and a joint schematic view are shown in Figure 4-1. The leg of the robot has six active degrees of freedom plus an optional passive degree of freedom in the foot. The vertical axis, Z, is the yaw axis. The X axis is the roll axis and the Y axis is the pitch axis. The hip has three degrees of freedom. These three degrees of freedom are made up of a universal joint (yaw and roll) followed by a pin joint (pitch). The pitch pin joint is oset slightly(about 2cm) from the yaw and roll axes. The frontal plane dimensions of the robot are shown in Figure 4-2. The dimensions are very close to the dimensions for a 50th percentile US male as given by Whitney(2). The ranges of motion are adapted from robot simulations and data found in Rose, et al.(18), Winter(20), and Kapandji(7). deg 80,-30 30,-20 30,-15 80,0 45,-20 20,-20 rad/s 7.3333 6.8 5.5 8.8 8.8 7.3 Nm 50 59 67 42 88* 100* Drive Type Pulley Push-rod Push-rod Push-rod Push-rod Push-rod

hip(pitch) hip(roll) hip(yaw) knee(pitch) ankle(pitch) ankle(roll)

Table 4.1: Robot Joint Specications. Torque and rad/s numbers are given for maximum moment arm. Power, torque, and velocity are symmetric due to the actuator. *The ankle roll and pitch are not independent. Their maximum values can not be applied simultaneously.

19

CHAPTER 4. ROBOT MECHANICAL DESIGN

20

Sagittal

Frontal

Transverse

Z - yaw

Y - pitch X - roll

Figure 4-1: A photo of the completed robot and a joint schematic view of the robot. The schematic shows active degrees of freedom only. The optional passive toe joint is not shown.

CHAPTER 4. ROBOT MECHANICAL DESIGN

21

60.00

16.00

7.25 (Hip Spacing) Hip Universal (Yaw, Roll)

40.00

37.75 37.00 Hip Pitch Axis

Knee Pitch Axis 20.00 4.50 Shin and Thigh Width Ankle Universal (Pitch, Roll)

4.00

3.00 0

Figure 4-2: The dimensions of the biped in the frontal plane. The body and links are approximately axially symmetric about their longitudinal axis. Dimensions are in inches.

CHAPTER 4. ROBOT MECHANICAL DESIGN Body Part/Area Shin & Foot (x2) Thigh (x2) Ab/Pelvic Arm (x2) Thorax to Head Human 6% 10% 27% 5% 31% Biped Robot 13.5% 11% 51 % NA NA PDW 10% 15% 50% NA NA

22

Table 4.2: Approximate distribution of mass in humans, the biped M2, and McGeers kneed passive dynamic walker(PDW). Human data adapted from Dempster and Gaughran (3). Robot weight distribution is driven primarily by actuator locations. Each actuator is approximately 1.2kg(2.5 lbs). The mass distribution of the robot is dominated by the location of the actuators within the links. As a result, the robots mass distribution is centered lower than an average humans. Table 4.2 shows the percentage mass distributions for an average male, M2, and a planar passive dynamic walker(10). The robot mass distribution is closer to that of a planar passive dynamic walker than to that of a human. Due to successful computer simulations(16), it was not necessary to add additional weight to the torso in order to put the proportions more in line with a human.

4.2.2

Actuators

The actuators used in the robot are 90W, 1.2KG Series Elastic Actuators(17). They are capable of a maximum force of roughly 1320N (300lbs) and a maximum speed of roughly 0.28 m/s (11 in/s). The actuators have a force control bandwidth of 30Hz. Linear actuators where chosen over rotary actuators due to the available space in the robot. Linear actuators allowed for placement along the longitudinal axis of the leg links. The actuators are symmetric in their power, speed, and force capabilities.

4.2.3

Ankle

The ankle of the biped is a universal joint. The pitch axis is followed by the roll axis. This is a slight deviation from the structure of the human ankle. The human ankle is often likened to a universal joint where the second axis is at 45 degrees to the rst rather than at 90 degrees(7). For engineering simplicity we use a universal joint with orthogonal axes. The instantaneous power requirement for ankle pitch is the greatest. The actuators were placed in a conguration so they can act together in the pitch direction. Photos of the ankle are shown in Figure 4-7 and a schematic of the ankle and its actuators are shown in Figure 4-3. The ankle has two series elastic actuators placed along the longitudinal axis of the shin. The actuators are mounted by a universal joint near the top of the shin and attached to the foot by a ball and socket joint (rod-end). This is a linkage variation of a standard geared dierential. When the actuators push in unison, a moment is generated about the pitch axis. When they push in opposite directions, a roll moment is generated.

4.2.4

Foot

Two dierent feet were designed for the robot. One foot is a simple rectangular design with four single axis load cells residing at each of the corners. This foot closely resembles the foot that was used in computer simulations in the lab. Another more involved foot was designed in order to explore the roll of the toes in walking.

CHAPTER 4. ROBOT MECHANICAL DESIGN

23

Actuators Y axis
E

X axis Ball Joint

O

Figure 4-3: A schematic of the ankle joint actuation scheme. The axes shown are xed to the center of the universal joint. The points A,E, and O are referenced in Appendix A.
4.00 .19 2.88
Toe Joint Axis

Heel

Ankle Joint

0 .50

2.13

6.69

Figure 4-4: The dimensions of the biped foot with passive toe joint. The Xs represent the location of the load cells. The more intricate second foot of the biped robot contains a passive joint which is modeled after the toe of a human. The joint is believed to smooth the center of mass trajectory of the body during a walking cycle(18). The toe joint is simply a pin joint with two limit stops and a soft return spring. Ground contact and sensing on the foot consists of four single axis load cells. One cell is placed at the heel and three cells are placed in a triangle at the toe/ball of the foot. The three cells in the toe are all constrained to the same plane. The three toe contact points can rotate about the foot Y-axis with respect to the heel contact point. The layout of the sensors can be seen in Figure 4-4. Standard through-hole button load cells are used. There is a rubber bumper which contacts the ground roughly 0.875 in diameter attached to each load cell. On the simple four point foot there is a rectangular pastic piece with a 0.25 piece of neoprene for grip and shock absorbtion. This can be seen in Figure 4-7. Single axis load cells were chosen over a six axis sensor because of their size and weight, and chosen over strain gauges because of their ease of use and quick replaceablility. On the more complicated biped foot, force control and a passive toe joint are the reasons the four contact points of the foot are not over-constrained. Since ankle roll is force controlled rather than position controlled, it can adjust itself so the three points of the toe lie at. The passive toe

9.09 9.59

CHAPTER 4. ROBOT MECHANICAL DESIGN

24

Mr
R O L L Yaw Body frame

Yaw

O
Roll Pitch

My Ar
Thigh

Figure 4-5: A schematic of the biped hip. The pitch actuator is not shown. It lies along the axis of the thigh.

Figure 4-6: A photo of the biped thigh illustrating the hip pitch cable drive system. joint(a pitch joint) then allows the fourth point on the heel to lie at as well.

4.2.5

Hip

The biped hip has three degrees of freedom. The joint consists of a universal joint followed by a slightly oset pin joint. A schematic of the hip is shown in Figure 4-5. The yaw axis is rst and the the yaw actuator is mounted to the body frame with a pin joint. The roll actuator is next. Since its attachment point passes through the roll angle, it is mounted to the body by a universal joint. Its endpoint is attached by a ball and socket joint. The pitch actuator, which is not shown, lies along the longitudinal axis of the thigh. It is the only actuator which is attached using a cable and pulley rather than a rod-end. The range of motion of the hip pitch joint is the largest. The moment arm changes associated with a drive arm would be too great at the extents of the pitch motion. At sixty degrees from the perpendicular drive position, the moment arm would be half its original length.

CHAPTER 4. ROBOT MECHANICAL DESIGN

25

45 deg 20 deg 20 deg

Figure 4-7: The ranges of motion on the robot ankle joint. The ankle roll is symmetric(+- 20 deg).

4.2.6

Knee

The robot knee is a simple pin joint. The joint is actuated by a SEA located in the thigh. Two photos demonstrating the range of motion of the knee can be seen in Figure 4-8.

CHAPTER 4. ROBOT MECHANICAL DESIGN

26

80 deg

Figure 4-8: The range of motion of the robot knee. The knee has 80 degrees of motion.

Chapter 5

Electronics and Control Systems

5.1 Electronics Overview

The basic electronic subsystems of the robot are shown in Figure 5-1. The robot is powered by a 48 Volt power supply. There is also an Ethernet connection not shown in the gure which can be used to load control code and retrieve data from the on-board computer. The circuit boards are summarized in Table 5.1. The sensors are summarized in Table 5.2. All signals are sent and received dierentially between the computer and the analog boards.

5.1.1

48 Volt bus

All the electronics on the robot receive power from a 48 Volt bus. The current setup has the robot connected to several oboard 48V power supplies. All the electronics were designed with this power bus in mind. The voltage choice was dictated by the brushless motor ampliers. It was also chosen because it is a standard voltage and the robot will eventually be augmented with batteries for future autonomous demos.

5.1.2

Robot Sensors

The robot is equipped with a total of 33 sensors. There are 12 rotary potentiometers that are used to measure joint position and this is dierentiated to give the joint velocity. The joint position measurements are relative displacements between joints. There are 12 linear potentiometers which measure the spring compression in the Series Elastic Actuators. These signals are local to the analog force control loop only. They are not used by the DSP in the control. There are 8 single axis load cells on the feet of the robot. These sensors are connected to the strain gauge conditioning board which amplies and buers the signal on its way to the DSP. There is a single Intersense Inertia

Table 5.1: Circuit Boards on the Robot QTY 12 1 1 1 6 2 4 1 Voltage Input +48 +5 5,12 5,12 +48 +48 12 +12 Vendor Copley Controls Dideas Dideas Dideas Leg Lab Leg Lab Leg Lab Intersense

Motor Ampliers DSP ANA074 ANA063 Analog Force Control Board Power Distribution Board Signal Conditioning Board IS-300 Signal Conditioner

27

CHAPTER 5. ELECTRONICS AND CONTROL SYSTEMS

28

Table 5.2: Sensors on the Robot Sensor Rotary Potentiometers Linear Potentiometers Single-Axis Load Cells 3-axis inclinometer QTY 12 12 8 1 Voltage Input +-5 +-5 +-9 NA Vendor Bourns Novotechnik Transducer Techniques Intersense

48Volt bus Vestibular Sensors

Brushless Motor Amp

Actuator

Computer A/D and D/A

Analog PID Regulated 5V Buffers Joint Pots

Instrumentation Power +-5 +-12

Load Cells

Figure 5-1: An overview of the biped electronics. Thick lines indicate power transfer and arrowheads indicate information ow.

CHAPTER 5. ELECTRONICS AND CONTROL SYSTEMS

29

Cube which measure the roll, pitch, and yaw(magnetic) of the robot body. The Inertia Cube is connected to a IS-300 motion tracker unit which communicates with the DSP through a RS-232 serial line. Table 5.2 provides a summary of the robot sensors. Figure 5-2 contains an annotated photo of M2 with the sensor locations.

Figure 5-2: A photo of M2 with annotations of electrical systems.

5.2
5.2.1

Custom Circuit Boards

The Signal Conditioning Boards

The signal conditioning boards are four layer printed circuit boards. They have a ground plane, a power plane, and two routing layers. The signal conditioning boards handle two 350 ohm wheatstone bridge sensors each. They have adjustable gain and osets.

5.2.2

The Analog Force Control Boards

The analog force control board provides two channels of PID force control for the Series Elastic Actuators and two channels of joint potentiometer buering and dierentiation. The analog force control boards have a 48 Volt power input and communicate with the DSP through dierential ampliers and receivers. The analog force control boards are four layer printed circuit boards. They have a ground plane, a power plane, and two routing layers.

CHAPTER 5. ELECTRONICS AND CONTROL SYSTEMS

30

Figure 5-3: A photo of the biped strain gauge signal conditioning board. This board handles two strain gauges. There are a total of four of these boards on the robot.

Figure 5-4: A photo of the analog force control and joint pot buer board. There are six identical boards on the robot.

CHAPTER 5. ELECTRONICS AND CONTROL SYSTEMS

31

5.2.3

The Analog Breakout Board

The analog breakout board connects to the ANA070 and ANA064 analog input and output boards. It has no power and simply allows for the use of smaller connectors to easily distribute the 64 channels of analog input and output from the ANA070 and ANA064 boards. There are two 64 pin connectors which connect to the ANA064 and ANA070 boards respectively. Then there are multiple 3,5, and 10 pin connectors which connect to the signal conditioning boards and analog force control boards. The analog breakout board is a two layer printed circuit board.

5.2.4

The Power Distribution Boards

The power distribution board has a 48 Volt input and outputs a variety of lower voltages. It appears in two places on the robot. There is one printed circuit for the two power distribution boards but there are two dierent DC-to-DC converters which are used when assembling the board. Board Conguration 1 This board supplies power to the DSP (+5V) and two analog boards (+-12V, +-5V). The analog and digital grounds are isolated. Board Conguration 2 This board provides power to the Intersense IS-300(+12V) and the four signal conditioning boards (+-12V). The power distribution boards are four layer printed circuit boards. They have a ground plane, a power plane, and two routing layers.

5.2.5

The DSP and Analog I/O

M2 has a custom DSP system that was designed by Chris Barnhart at Digital Designs and Systems. The DSP is a Texas Instruments C-31. There are three circuits boards that make up the system. 1. DSP. Mulitple Serial Lines and Digital I/O. 2. 16 Analog Ins, 16 Analog Outs, and Ethernet. 3. 32 Analog Inputs.

5.3

Control Software

Control code for the robot is written in C. It is compiled oine on a Unix machine and downloaded to the robot via Ethernet. The main robot control code runs at 500Hz on the DSP and sends desired forces to the analog boards. Currently, the control system treats the actuator and analog boards as a black box force source. This means it is assumed that the desired force sent out to the actuator is the actual force applied to the joint. This is true within the realm of frequencies needed for walking(less than 15Hz). The analog boards have PD control loops for the Series Elastic Actuators. The analog boards also buer and dierentiate the joint potentiometers and send the information to the DSP. This is similar to the electronics used for Spring Flamingo. The main dierences are the addition of Ethernet for faster communication with the outside world and dierential send and receive for signals within the robot.

Bibliography
[1] Jesper Adolfsson, Harry Dankowicz, and Arne Nordmark. 3-d stable gait in passive bipedal mechanisms. Proceedings of 357 Euromech, 1998. [2] Henry Dreyfus Associates. The Measure of Man and Woman. Whitney Library of Design, New York, 1993. [3] W.T. Dempster and G. Gaughran. Properties of body segments based on size and weight. American Journal of Anatomy, 1965. [4] J. Fowble and A. Kuo. Stability and control of passive locomotion in 3d. Proceedings of the Conference on Biomechanics and Neural Control of Movement, pages 2829, 1996. [5] Mariano Garcia, Anindya Chatterjee, and Andy Ruina. Speed, eciency, and stability of small-slope 2d passive dynamic bipedal walking. IEEE International Conference on Robotics and Automation, pages 23512356, 1998. [6] K. Hirai, M. Hirose, Y. Haikawa, and T. Takenaka. The development of honda humanoid robot. IEEE International Conference on Robotics and Automation, 1998. [7] Ibrahim Adalbert Kapandji. The Physiology of the Joints. Churchill Livingstone, 1982. [8] Benjamin Krupp. Design and control of a planar robot to study quadrupedal locomotion. Masters thesis, Massachusetts Institute of Technology, August 2000. [9] Tad McGeer. Passive dynamic walking. International Journal of Robotics Research, 9(2):6282, 1990. [10] Tad McGeer. Passive dynamic biped catalogue. Symposium of Experimental Robotics, 1991. [11] Thomas A. McMahon. Mechanics of locomotion. Reasearch, 3(2):428, 1984. Proceedings of the 2nd International The International Journal of Robotics

[12] Gill A. Pratt and Matthew M. Williamson. Series elastic actuators. IEEE International Conference on Intelligent Robots and Systems, 1:399406, 1995. [13] J. Pratt, P. Dilworth, and G. Pratt. Virtual model control of a bipedal walking robot. IEEE International Conference on Robotics and Automation, pages 193198, 1997. [14] Jerry E. Pratt. Virtual model control of a biped walking robot. Masters thesis, Massachusetts Institute of Technology, August 1995. [15] Jerry E. Pratt and Gill A. Pratt. Exploiting natural dynamics in the control of a planar bipedal walking robot. Proceedings of the Thirty-Sixth Annual Allerton Conference on Communication, Control, and Computing, pages 739748, 1998. [16] Jerry E. Pratt and Gill A. Pratt. Exploiting natural dynamics in the control of a 3d bipedal walking simulation. Proceedings of the International Conference on Climbing and Walking Robots (CLAWAR99), 1999. 32

BIBLIOGRAPHY

33

[17] David W. Robinson, Jerry E. Pratt, Daniel J. Paluska, and Gill A. Pratt. Series elastic actuator development for a biomimetic robot. IEEE/ASME International Conference on Advanced Intelligent Mechatronics, 1999. [18] Jessica Rose and James G. Gamble. Human Walking. Williams and Wilkins, 1994. [19] Mark Roshiem. Robot Wrist Actuators. John Wiley and Sons, Inc., 1989. [20] D. A. Winter. Biomechanics and Motor Control of Human Movement. John Wiley and Sons, Inc., New York, 1990. [21] Jinichi Yamaguchi, Eiji Soga, Sadatoshi Inoue, and Atsuo Takanishi. Development of a bipedal humanoid robot - control method of whole body cooperative dynamic biped walking. IEEE International Conference on Robotics and Automation, pages 368374, 1999.

Appendix A

Joint Math
Since all of M2s rotary joints are actuated by linear actuators, transformations are needed to get from joint torques to actuator forces. These are dependent upon the joint layout and actuation scheme. If the force is known, the torque is somewhat easily computed using the formula, = r F . The cross product = r F operation can be rewritten as a matrix multiplication = AF where A is a singular matrix. i j k 0 r3 r2 0 r1 F = AF = r F = r1 r2 r3 = r3 (A.1) F1 F2 F3 r2 r1 0 However, if the cross-product is written as = |R||F | sin() then we can easily invert. In the case of most of the robot joints, this requires use of the arctan function or the law of cosines to nd the angle . In general, the direction of F is xed by the geometry of the joint and |F | is the quantity of real interest.

A.1

Knee Joint

The knee joint is a single degree of freedom rotary joint allowing relative motion between the shin and the thigh. When all the joints of the robot are zero(it is standing with locked knees), the knee axis is parallel to the global Y axis. Rotation about the knee joint is referred to as pitch. The knee actuator is connected to the knee joint via a push rod of length |OA| = rk at the shin and a xed pin joint at the thigh. The knee joint has three points of interest which we will use for the derivation of the transformation. The knee pivot O, the actuator pushrod attachment A, and the actuator mounting pivot M . The robot knee joint and points can be seen in Figure A-1 and a simple line drawing is seen in Figure A-2. = r F = rk sin( OAM )Fknee The actuator force required given k is Fknee = k rk sin( OAM ) (A.3) (A.2)

where Fknee is a scalar and OAM can be dened as follows OAM = f ixed + k AM = f ixed + k arctan r sin(f ixed + k ) L2 L1 + r sin(f ixed + k ) (A.4)

34

APPENDIX A. JOINT MATH

35

Figure A-1: The robot knee joint with superimposed lines and points. Points O, A and M all refer to pin joints. O is the knee joint. M is where the actuator is mounted to the thigh and A is where the actuator is attached to the shin.

L1 L2

knee M MA A fixed

Figure A-2: A line drawing of the knee for the calculation of knee actuator desired force. The actuator is the line segment M A. This drawing also pertains to the geometry of the hip and ankle joints but the hip and ankle actuators have motions out of the plane whereas the knee actuator is always in the plane perpendicular to the knee axis.

APPENDIX A. JOINT MATH

36

O r

rx

A2

ry

A1

Figure A-3: A simple view of the actuation of the robot ankle. This is a tranverse plane slice of the robot ankle. Point O is the intersection of the X and Y axes as well as the center of the ankle universal joint. Points A1 and A2 represent the attachment points of the two actuators respectively. Forces in the actuators create torques about both the X and Y axes. Joint rotations change the relative lengths of the X and Y projections(rx and ry ) of the moment arms, therefore aecting the torque to force transformation. The angle between the shin and the thigh is k and the constant, f ixed , is the angle between the shin and the segment OA. It is also possible to derive the equations avoiding the inverse tangent by using the law of cosines. It would be possible to simplify the equations a bit by assuming the actuator is always parallel to the thigh, i.e. AM = 0. This simplies the equations and shouldnt reduce accuracy much considering that L1 L2 . Equation A.3 then becomes Fknee = k rk sin(f ixed + k ) (A.5)

where Fknee is once again a scalar quantity.

A.2

Ankle Joint

The ankle joint is a universal joint(2 d.o.f.) allowing relative motion between the foot and the shin. When all of the joints of the robot are zero, the axes of the ankle joint line up with the global X and Y axes. The Y axis is proximal to the shin. Rotation about the Y axis is referred to as pitch and rotation about the X axis is referred to as roll. The math for the ankle is an extension of the math for the knee. We can consider the ankle as two decoupled cases of the knee joint. Then one can simply add the forces from aroll and apitch . There is a slightly more complicated analogy to the knee in two planes of the ankle. The ankle roll considers what is happening in the XZ plane and the ankle pitch considers what is happening in the YZ plane. Both planes are coming out of the page in Figure A-3. Each plane looks similar to the knee diagram shown in Figure A-2. The main dierence is that |OA| = r is a function of the other ankle angle.

APPENDIX A. JOINT MATH

37

Mr
R O L L Yaw Body frame

Yaw

O
Roll Pitch

My Ar
Thigh

Figure A-4: The robot hip joint schematic. Point O refers to the hip universal joint. The Z(hip yaw) and Y(hip roll) axes intersect at point O. Point Mr refers to the universal joint on the body where the roll actuator is mounted. Point Ar refers to the ball joint on the thigh where the roll actuator is attached. Point My refers to the pin joint where the yaw actuator is attached to body. Point Ay is the pin joint attachment of the yaw actuator to the yaw universal block. It is hidden in this schematic.

A.3
A.3.1

Hip Joint
Hip Pitch

The math for the hip pitch is the simplest of all the joints since the hip pitch actuator is attached by a pulley. The desired torque at the joint(hp ), leads to the force(fhp ) command to the hip pitch linear actuator. fhp = hp rhippitch (A.6)

The transformation from actuator velocity to joint velocity is quite simple as well. velhp 2rhp

hp =

(A.7)

A.3.2

Hip Roll and Yaw

The yaw actuator transformation is identical to that of the knee actuator except for the addition of a small term which can be ignored with little aect. The actuator force required given hyaw is Fhyaw = hyaw + sin( Oy Ay My )
hroll

(A.8)

where is a small factor due to the torque applied to hip roll which is a function of the hip roll and yaw angles. Oy , Ay , and My are dened as they were in the case of the knee.

APPENDIX A. JOINT MATH

38

The actuator force required given hroll can be simplied to Fhroll = hroll (1 + sin( Ohr Ahr Mhr )
hyaw )

(A.9)

The exact equation for the hip roll incorporates the yaw angle. Since the hip roll actuator attachment point Ahr is attached to the leg below the yaw degree of freedom, the points Ohr , Ahr , and Mhr do not always lie within the plane perpendicular to the roll axis as in the case of the knee joint where they are xed in the plane perpendicular to the knee axis. When the roll actuation plane(plane formed by the three points Ohr , Ahr , Mhr ) is skewed with respect to the roll axis, a small torque will be contributed to the yaw axis. This is where the term hroll comes from in Equation A.8 and this is why the hip roll equation has the hyaw term in it.

Appendix B

Electrical Schematics
This Appendix contains the electrical schematics for all the custom circuit boards used on the robot. The basic functionality of the boards and their role in the overall robot is decribed in more detail in Chapter 5. There are six pages for the analog force board, 2 pages for the signal conditioning board, 1 page for the analog breakout board, and 1 page for the power distribution board.

39

TP1 PB1 plus12 1 1 2 +INPUT 1uf 2 1 Local_GND 1uf -12 2 12GND GND2 GND3 GND2 GND4 2 Actuator Command from DSP minus12 CA+ CACB+ JJP1 +5 1 +5 +5 1 +5 Signal -5 4 GND 4pin_pot FPB +5 1 Force_Pot_B 2 -5 3 4 JJP2 +5 1 +5 Signal -5 4 3 -5 5GND 4pin_pot Copley_B+ CCB+ IN 5PIN_copleystyle 2 PC3 CCBCopley_BA D

Test Point +12OUT 1 12+ 3 BC1 4 BC2 5 +12

J1 1 220uf 2 -12OUT 2 -INPUT

+48 GND PC1

48IN_CONN

DC/DC CONVERTER CMN +-12V

Test Point Test Point Test Point GND0 Test Point

5+ DFC10U48D12 TP3 3 Test Point plus5 1 TP2 Test Point +5 GND1

REG1

GND20

+12

IN

+5OUT

GND

J2 Command_In_A+ Command_In_ACommand_In_B+ CBCommand_In_B1 2 3 4 +Vo A -Vo A +Vo B -Vo B 4pin

C

I43406 2 2 Joint Pot Signals out to DSP 2 Joint Pot Derivatives out to DSP JFP1 Signal FPA 2 3 -5 Force_Pot_A -5 GND 4pin_pot J3 Joint_Pot_A 2 3 4 -5 JPA

PC2 10uf

Joint_Pot_A+_out

JPA+

APPENDIX B. ELECTRICAL SCHEMATICS

JPA-

Joint_Pot_A-_out

Local_GND

Joint_Vel_A+_out

JFP2 +5 Signal -5 GND 4pin_pot CCA+ Copley_A+ Copley_ACCALocal_GND 2 Current Commands on their way to Copleys J4
B

JVA+

JVA-

Joint_Vel_A-_out

1 2 3 4 5 A+ AGND AV+ AV-

JPB+

Joint_Pot_B+_out

JPB-

Joint_Pot_B-_out

Local_GND 10PIN_copleystyle JPB -5 3 Joint_Pot_B 2

JVB+

Joint_Vel_B+_out

6 7 8 9 10 B+ BGND2 BV+ BV-

Test Point

Figure B-1: Page 1 of 6 of the biped force board.

1 2 3 4 5 I43314 10uf TP4 minus5 Size A Date:
4 3

JVB-

Joint_Vel_B-_out

REG2

GND -5OUT

A+ AGND B+ B-

-12

Title MIT LEG LAB, BIPED FORCE BOARD Document Number Rev {RevCode}

POWER AND CONNECTORS

Tuesday, December 21, 1999
2

Sheet

1
1

of

40

TP5 FC1 5k 0.1uf R2 5k 0.1uf TP6 BC3 U1B 7 LM324 12R6 TP7 U1D14 LM324 12+ 1 BC5 R9 5k 5k Local_GND 1 0.1uf 2 BC6 1211 FEB 3 2 U2A 1 + LM324 4
B

R1

11

Test Point 1 U1A 1 11 LM324 6 + 5 2 5k R3 Local_GND

5+ + 4 R5 11 8 LM324 13 + TP8 5k 12 4 1 Test Point 5k 4 Test Point FEA RCV1 2 1uf FA_DES 2 R8 Sense A -In A Out A +IN A Ref A G=1/2 6 -In B Out B +In B Ref B VFB_DES INA2137 BC7 FC2 0.1uf TP9 R13 4 1 5k U2B 5k Local_GND TP11 Test Point 10 TP12 11 Test Point U2D 14 LM324 5k Size A Date:
4 3

FPA 3

R4

10k 4

POT1 10 + 11 9

U1C

20k

R7 12+ 11 BC4 Test Point

10k

5V+

CA2 13 14 10 R10 9 8 5k 3

APPENDIX B. ELECTRICAL SCHEMATICS

CA+

CB5

CB+

5+ 4

12-

R11 10k 1 1uf Local_GND

Test Point

6 11 11 LM324 9 5k

R15

10k R16

513 + 12 4

FPB

Figure B-2: Page 2 of 6 the biped force board.

12 0.1uf Sense B R12 12+ 5 + 7 R14 BC8 0.1uf U2C 8 LM324 Title MIT LEG LAB, BIPED FORCE BOARD Document Number

POT2 20k

TP10

Test Point

FORCE ERROR, A and B

Wednesday, December 15, 1999
2

Rev {RevCode} Sheet 2

1

of

41

R17 FA_DES 5k RFF1 11 R18 1 LM324_0 357 ohm 4 Test Point R20 .01uf R21 40.2 k U3B 7 LM324_0 4.42 K 9 10 R23 11 R22 6 1uf 4 + 402 ohm 5 C2 11 10k U3C 8 + LM324_0 4 6 R24 4 open BC10 12- 2 1 0.1uf U3D 14 LM324_0 BC12 2 0.1uf Title MIT LEG LAB, BIPED FORCE BOARD Size A Date:
4 3 A

2 + 5k 3

U3A

R19

Open

FEA

C1

TP13

APPENDIX B. ELECTRICAL SCHEMATICS

PID_A

12+ BC9 C3 1 2 0.1uf DRV1 V+ +Vo Vin 3 Gnd V-Vo 5 +Sense -Sense 8 7 2 1 CCAR26 12open CCA+
B

R25 13 + open 12 4

11

Figure B-3: Page 3 of 6 of the biped force board.

open DRV134 BC11 2 1 0.1uf 12+ 1 Document Number

PID A
Wednesday, December 15, 1999
2

Rev {RevCode} Sheet 3

1

of

42

R27 FB_DES 5k RFF2 11 R28 5k + 3 LM324_0 357 ohm 4 1 C4 TP14 Test Point R30 .01uf R31 40.2k U4B LM324_0 + 10 4 4.42k 7 8 LM324_0 9 U4C R33 11 R32 6 1uf 4 + 402 ohm 5 C5 11 10k PID_B
C

U4A

R29

open
D

FEB

APPENDIX B. ELECTRICAL SCHEMATICS

12+ BC13 1 6 C6 open R34 open BC14 12- 2 1 0.1uf U4D 14 LM324_0 open R36 12R35 13 + open 12 4 11 2 0.1uf DRV2 V+ +Vo 4 3 Vin Gnd V-Vo 5 DRV134 BC15 2 1 0.1uf +Sense -Sense 8 7 2 1 CCBCCB+
B

Figure B-4: Page 4 of 6 the biped force board.

12+ 1 2 BC16 0.1uf Title MIT LEG LAB, BIPED FORCE BOARD Size A Date:
4 3

Document Number

PID B
Wednesday, December 15, 1999
2

Rev {RevCode} Sheet 4

1

of

43

BC17 12+ 1 4 11 5k JPA_BUF V+ +Vo JPA+

D

C7 BC18 2 TP15 DRV3 8 7 2 1 JPATest Point 7 Vin +Sense -Sense Gnd V-Vo DRV134 12LM324 3 4 6 5 2 0.1uf DRV4 +Vo +Sense -Sense Gnd V-Vo DRV134 5 2 1212+ JPB_BUF 4 Local_GND 4 3 BC26 1 6 2 0.1uf 12+ DRV6 V+ +Vo 4 3 LM324 TP18 4 Vin 8 Gnd V-Vo 5 BC28 1 0.1uf 2 DRV134 12A

2 0.1uf U5A LM324 + 5 4 10k 1 6 U5B R38 R37 0.1uf 0.033uf 12+

JPA 3 + 11 12C8 BC20 0.33uf 11 R40 LED R41 LM324 3 4 D2 1k C10 BC23 12+ 0.033uf R42 R43 5k U6B 7 LM324 VTP17 C11 0.33uf R44 R45 RL3 9 D3 LED R46 14 LM324 1k D4 LED Title + 1.65 k 1uF 10 50k U6D U6C C12 11 5k Gnd -Vo 5 DRV134 BC27 128 +Sense -Sense 7 2 1 JVBJVB+ 2 1 0.1uf 1 + LM324 10k 5 6 11 4 0.1uf U6A 1 2 TP16 1 0.1uf LED BC22 + 1.65 k 10 1uF D1 9 8 Vin 4 U5C V+ 8 7 2 1 JVAC9 5k 6 R39 0.1uf 2 12+ 1 1 BC21 2 BC19 1 2 0.1uf

JPA_BUF

RL2

RL1 5k

50k

JVA+

Test Point

APPENDIX B. ELECTRICAL SCHEMATICS

V+

JPB 3 + 11 2

Test Point

12

13

11

Figure B-5: Page 5 of 6 of the biped force board.

1 Vin 12BC25 0.1uf Test Point Size A Date:
4 3

JPB_BUF

RL4

5k

12

Document Number
2

13

11

LM324 14

U5D

BC24 2 0.1uf DRV5 +Vo +Sense -Sense 8 7 2 1 JPBB

JPB+

MIT LEG LAB, BIPED FORCE BOARD

JOINT POSITION AND VELOCITY
Wednesday, December 15, 1999 Sheet 5
1

Rev {RevCode} of 6

44

RL5

50K D5 FA_DES JMP1 JUMPER RL7 D6 4 LED + 3 U7A 500ohm 5k RL9 open D9 JMP3 FEA JUMPER 5k RL10 9 10 11 LED LM324 8 + U7C 4 R49 1k D10 LED
B

RL6 2 LM324 1 R47

11

LED
D

50K BC29 12+ 1 D7 LED R48

C

2 0.1uf LM324 7 U7B 500ohm D8 LED BC30 2 0.1uf 1

FB_DES 5k + 5 4 12RL11 open D11 JMP4 5k + 12 U7D 4 13 JUMPER 1k D12 LED Title LM324 14 R50 RL12 11 LED Size A Date:
4 3

JMP2 6

RL8

APPENDIX B. ELECTRICAL SCHEMATICS

JUMPER

Figure B-6: Page 6 of 6 the biped force board.

11 Document Number LED MONITORS Wednesday, December 15, 1999
2

FEB

MIT LEG LABORATORY, BIPED FORCE BOARD Rev {RevCode} Sheet 6

1

of

45

J1 +12V U1A 11 LF347 1 1 2 Strain GND STRAIN_CONN Strain 1 J2

D

Vexc_out Sense-High +Input -Input Sense-Low GND BRIDGE_CONN 4 J3 U2 Strain 2 -12V J4 1 2 Strain GND STRAIN_CONN
C

1 2 3 4 5 6 + 1 2 3 4 5 6 Vexc_out 2 Sense-High 2 +Input 2 -Input 2 Sense-Low 2 TP1 TP2 TP3 TP4 TP5 TP6 TP7 TP8 TP9 Test Point Test Point Test Point Test Point Test Point Test Point Test Point Test Point Strain 1 Strain 2 +12V +9V reg -9V reg -12V +12V +9V reg Test Point R4 620 2 2 D1 LED -12V R5 470 2 D2 LED R6 470 D3 LED 2 R7 620 D4 LED 1 1 1 -9V reg -12V 1 3 2

Vexc_out 1 Sense-High 1 +Input 1 -Input 1 Sense-Low 1

For Half-Bridge Completion: 1 - Connect pin 2 and pin 25 2 - Leave connector -Input open

Strain Gain Adjust

R1 Strain Gain Adjust 1 C1 4.7uF BRIDGE_CONN +12V R2 15k

1 2 3 4 + Input - Input Gain 1 Gain 2 Vexc Out Sense High Sense Low Half-Bridge Completion

28 27 26 25

Vexc_out Sense-High +Input -Input Sense-Low GND

+9V reg

Strain Gage Signal Conditioner

Vout Unfiltered Input Offset Adj 1 Input Offset Adj 2 Output Offset Adj Bandwidth Adj. 1 Bandwidth Adj. 2 Vout Filtered 1B31 C2 1uF C4 1000pF -12V 1000pF C5 1uF C3 Exc. Adj Ref. In Ref. Out + Vs Reg + Vs Comm - Vs 21 20 19 18 17 16 15

APPENDIX B. ELECTRICAL SCHEMATICS

R3 50k

8 9 10 11 12 13 14

-9V reg

+12V +9V reg J5 +9OUT U4 C7 IN I02162 0.1uF -9V reg 1 GND -9OUT 3 GND 3 0.1uF C6

+12V

U3

IN

+12V GND -12V FLAM_PWR_CONN

1 2 3

Figure B-7: Page 1 of 2 of the strain gauge conditioning board.

2 I02164 Title -12V Size A Date:
4 3

MIT Leg Laboratory

M2/M4 Protoptype Strain Gage Board

Document Number Friday, August 14, 1998
2

Rev Sheet 1
1

of

46

+12V U1B 11 LF347 7 Strain 2

For Half-Bridge Completion: 1 - Connect pin 2 and pin 25 2 - Leave connector -Input open U5 +Input 2 -Input 2 + Input - Input Gain 1 Gain 2 +12V C8 11 4.7uF R9 15k +12V 9 10 Vexc Out Sense High Sense Low Half-Bridge Completion Vexc_out 2 Sense-High 2 Sense-Low 2 -12V 4 28 27 26 25 U1C LF347 8 + + 5

Strain Gain Adjust

R8 Strain Gain Adjust 2

1 2 3 4

+9V reg

Strain Gage Signal Conditioner

APPENDIX B. ELECTRICAL SCHEMATICS

R10 50k 2

8 9 10 11 12 13 14 Vout Unfiltered Input Offset Adj 1 Input Offset Adj 2 Output Offset Adj Bandwidth Adj. 1 Bandwidth Adj. 2 Vout Filtered 1B31 Exc. Adj Ref. In Ref. Out + Vs Reg + Vs Comm - Vs

21 20 19 18 17 16 15

-9V reg -12V

-12V 11

Figure B-8: Page 2 of 2 of the strain gauge conditioning board.

-12V 4 Title M2/M4 Prototype Strain Gage Board Size A Date:
4 3

Document Number {Doc} Friday, August 14, 1998

2

Sheet

12

13 2

+12V U1D LF347 14

B

Rev 1 of
1

47

IN-R-FOOT1 JP2 1 2 3 4 5 A+ AGND B+ B5PIN_copleystyle IN-X2 ADC+34 ADC-34 GND ADC+35 ADC-35 1 2 3 4 5 A+ AGND B+ B5PIN_copleystyle IN-X7 ADC+44 ADC-44 GND ADC+45 ADC-45 1 2 3 4 5 A+ AGND B+ B5PIN_copleystyle OUT-THIGH1 DAC4+ DAC4GND DAC5+ DAC51 2 3 4 5 A+ AGND B+ B5PIN_copleystyle OUT-BODY2 GND3 Test Point GND1 5PIN_copleystyle IN-THIGH2 ADC16+ ADC161 2 3 4 5 ADC14+ ADC14ADC15+ ADC15A+ AGND AV+ AVADC17+ ADC17ADC18+ ADC186 7 8 9 10 B+ BGND2 BV+ BVB+ BGND2 BV+ BV10PIN_copleystyle OUT-X4 Title 1 2 3 sig+ sigGND 3pin_copley_style
3

IN-R-FOOT2 ADC26+ ADC26GND ADC27+ ADC271 2 3 4 5 A+ AGND B+ B5PIN_copleystyle IN-X3 ADC+36 ADC-36 GND ADC+37 ADC-37 1 2 3 4 5 A+ AGND B+ BA+ AGND B+ B1 2 3 4 5 ADC+38 ADC-38 GND ADC+39 ADC-39 IN-X4 5PIN_copleystyle
D

IN-L-FOOT1 ADC28+ ADC28GND ADC29+ ADC291 2 3 4 5 A+ AGND B+ BA+ AGND B+ B1 2 3 4 5 ADC30+ ADC30GND ADC31+ ADC31-

IN-L-FOOT2

JP1

ADC24+ ADC24GND ADC25+ ADC255PIN_copleystyle IN-X1 ADC+32 ADC-32 GND ADC+33 ADC-33 1 2 3 4 5 A+ AGND B+ B5PIN_copleystyle IN-X6 ADC+42 1 ADC-42 2 GND 3 ADC+43 4 ADC-43 5 A+ AGND B+ B5PIN_copleystyle OUT-SHIN2 DAC2+ DAC2GND DAC3+ DAC31 2 3 4 5 A+ AGND B+ B5PIN_copleystyle 5PIN_copleystyle IN-X5 ADC+40 ADC-40 GND ADC+41 ADC-41 1 2 3 4 5 A+ AGND B+ B5PIN_copleystyle OUT-SHIN1 DAC0+ DAC0GND DAC1+ DAC11 2 3 4 5 A+ AGND B+ B5PIN_copleystyle OUT-BODY1

5PIN_copleystyle IN-X8 ADC+46 ADC-46 GND ADC+47 ADC-47 1 2 3 4 5 A+ AGND B+ B5PIN_copleystyle

C

OUT-THIGH2 DAC6+ DAC6GND DAC7+ DAC71 2 3 4 5 A+ AGND B+ B5PIN_copleystyle

APPENDIX B. ELECTRICAL SCHEMATICS

DAC0+ DAC1+ DAC2+ DAC3+ DAC4+ DAC5+ DAC6+ DAC7+ DAC8+ DAC9+ DAC10+ DAC11+ DAC12+ DAC13+ DAC14+ DAC15+ ADC+32 ADC+33 ADC+34 ADC+35 ADC+36 ADC+37 ADC+38 ADC+39 ADC+40 ADC+41 ADC+42 ADC+43 ADC+44 ADC+45 ADC+46 ADC+47 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52 54 56 58 60 62 64 HEADER64 GND2 Test Point GND0 IN-SHIN1 ADC4+ ADC4ADC8+ ADC8ADC9+ ADC9ADC10+ ADC10ADC11+ ADC116 7 8 9 10 10PIN_copleystyle OUT-X3 DAC14+ DAC14DAC15+ DAC151 2 3 1 2 3 4 5 ADC5+ ADC5ADC6+ ADC6GND DAC7+ ADC710PIN_copleystyle OUT-X2 DAC13+ DAC131 2 3 sig+ sigGND 3pin_copley_style
4

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52 54 56 58 60 62 64

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63

2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52 54 56 58 60 62 64

DAC0DAC1DAC2DAC3DAC4DAC5DAC6DAC7DAC8DAC9DAC10DAC11DAC12DAC13DAC14DAC15ADC-32 ADC-33 ADC-34 ADC-35 ADC-36 ADC-37 ADC-38 ADC-39 ADC-40 ADC-41 ADC-42 ADC-43 ADC-44 ADC-45 ADC-46 ADC-47 ADC31+ ADC30+ ADC29+ ADC28+ ADC27+ ADC26+ ADC25+ ADC24+ ADC23+ ADC22+ ADC21+ ADC20+ ADC19+ ADC18+ ADC17+ ADC16+ ADC15+ ADC14+ ADC13+ ADC12+ ADC11+ ADC10+ ADC9+ ADC8+ ADC7+ ADC6+ ADC5+ ADC4+ ADC3+ ADC2+ ADC1+ ADC0+ 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52 54 56 58 60 62 64 ADC31ADC30ADC29ADC28ADC27ADC26ADC25 ADC24ADC23ADC22ADC21ADC20ADC19ADC18ADC17ADC16ADC15ADC14ADC13ADC12ADC11ADC10ADC9 ADC8ADC7ADC6ADC5ADC4ADC3 ADC2ADC1ADC0-

HEADER64 GND1 Test Point GND IN-SHIN2 DAC8+ DAC8GND DAC9+ DAC91 2 3 4 5 A+ AGND B+ B-

DAC10+ DAC10GND DAC11+ DAC11-

1 2 3 4 5

A+ AGND B+ BIN-BODY1

IN-BODY2 ADC20+ ADC201 2 3 4 5 A+ AGND AV+ AVADC21+ ADC21ADC22+ ADC22ADC19+ ADC196 7 8 9 10 B+ BGND2 BV+ BV10PIN_copleystyle ADC23+ ADC23-

ADC0+ ADC0A+ AGND AV+ AV1 2 3 4 5 A+ AGND AV+ AV-

ADC1+ ADC1-

1 2 3 4 5

5PIN_copleystyle IN-THIGH1 ADC12+ A+ ADC12AGND ADC13+ AV+ ADC13AV-

1 2 3 4 5

A+ AGND AV+ AV-

ADC2+ ADC2B+ BGND2 BV+ BV6 7 8 9 10 B+ BGND2 BV+ BV-

ADC3+ ADC3-

6 7 8 9 10

6 7 8 9 10

B+ BGND2 BV+ BV10PIN_copleystyle

A

10PIN_copleystyle OUT-X1

DAC12+ 1 DAC12- 2 3

sig+ sigGND

sig+ sigGND 3pin_copley_style

M2 BREAKOUT BOARD Size A Date: Document Number {Doc} Monday, March 06, 2000
2

3pin_copley_style

Rev {RevCode} Sheet 1

1

of

48

Figure B-9: Page 1 of 1 of the analog I/O breakout board. This board is designed to interface with the ANA070 and ANA064 from Digital Designs and Systems.

J1 1 PB1 TP1 plus12 J2 +12 1 +V GND -V

D

+48 GND 1 PC1 2 48+ 1 +INPUT -INPUT -12OUT 2 +12 minus12 2 DFC10U48D12 -12 3 -V Second power brick is either +5 or +12 single ended supply. GND2 GND Test Point 2 ground 10uf PC4 TP3 Test Point plus 5 or 12 U3 J4 2 1 GND +V 2pin_terminal_block_ed1609 J5 5+ J6 REG1 TP4 3 +5 Test Point plus5 1 -5 2pin_terminal_block_ed1609 +12 1 IN GND +5OUT +5 1 2 3 +V GND -V 3pin_terminal_block_ed1610 PC5 10uf 2 +5 1 2 -5 REG2 1 GND -5OUT IN 3 -5 53 J7 +V GND -V 3pin_terminal_block_ed1610
B

1 2 +12 Test Point GND1 2 -12 ground0 3 Test Point TP2 -12 Test Point GND PC2 10uf 12 PC3 10uf +12OUT 4 5 3pin_terminal_block_ed1610 J3 1 +V 3

48+ 48-

48IN_CONN 482

DC/DC CONVERTER CMN +-12V

PB2

3pin_terminal_block_ed1610

5 U1 +12 -12 1 2 3 +5 out -5 3pin 4 1 3 +V

U2 +12 -12 1 2 3 +5 out -5 3pin

C

48no connection +5OUT

-INPUT

48+

+INPUT

DFA6U48S5

APPENDIX B. ELECTRICAL SCHEMATICS

+V

+V

2 1 GND +V

10uf

Test Point

Figure B-10: Page 1 of 1 of the M2 power distribution board. This board has two dierent population options. One to power the DSP and one to power the Intersense IS-300 tracker.
DC/DC CONVERTER +5V
+12 -12 1 2 3 +5 out -5 3pin I43406 I43314 -12 PC6 MIT Leg Laboratory Title TP5 minus5 Size A Date:
4 3

U4 +12 -12 1 2 3 +5 out -5 3pin

Power Board, February 2000

Document Number {Doc} Tuesday, February 15, 2000
2

Rev {RevCode} Sheet 1

1

of

49

Appendix C

Suppliers and Costs

This section contains a list of all the suppliers that were used in development of the robot. Digital Designs and Systems 3266 North Meridian Street Indianapolis, IN 46208 317-931-8190 http://www.dideas.com/ microcontrollers, a/d and d/a conversion, vision systems, ready built and custom designs Stock Drive Products 2101 Jericho Turnpike, Box 5416 New Hyde Park, NY 11042-5416 (516)328-3300 Mechanical Parts Berg 499 Ocean Avenue Box 130 East Rockaway, NY 11518 (516)596-1700 Mechanical Parts NOVOTECHNIK 237 CEDAR HILL STREET MARLBORO, MA 01752 (508) 485-2244 Potentiometers SMALL PARTS INC 13980 NW 58TH COURT PO BOX 4650 MIAMI LAKES, FL 33014-3115 (800) 220-4242 Mechanical Parts CENTURY SPRING COMPANY INC BOX 15287 222 EAST 16TH STREET

50

APPENDIX C. SUPPLIERS AND COSTS

51

LOS ANGELES, CA 90015-0287 (800)237-5225 Springs Newark Electronics 41 Pleasant Street Methuen, MA 01844 (800) 463-9275 Electronic Parts REC ENGINEERING COMPANY(Bob) 20 HOPKINS STREET WILMINGTON, MA 01887-2210 (508)657-6517 Machining Eastern Tool(Joe) Somerville, MA (617)497-6703 Machining Transducer Techniques 43178 Business Pk. Dr. Temecula CA 92590 1-800-344-3965 Load cells. Intersense, Inc. 73 Second Avenue Burlington MA 01803 (781)-270-0090 Multi-axis inclinometer. Digi-Key Corporation 701 Brooks Ave. South Thief River Falls, MN 56701-0677 (800)344-4539 Electronic Parts Copley Controls Corporation 410 University Avenue Westwood, MA 02090 (617)329-8200 FAX: (617)329-4055 Ampliers Cooner Wire Company 9265 Owensmouth Chatsworth, CA 91311 (818)882-8311 Flexible Wire

APPENDIX C. SUPPLIERS AND COSTS

52

CFC, Inc. 179 Bear Hill Road Waltham, MA 02154-1001 (617)890-1878 FAX: (617)890-7098 MODEM: (617)890-7193 Circuit Board Fabrication Advanced Circuits Colorado, http://www.4pcb.com Circuit Board Fabrication, online ordering NSK -linear motion division Elmhurst, IL 60126-1016 ATTN: Sandy x2638 or Jackie x2677 tel:1-800-255-4773 fax:630-924-8197 Ballscrews Litton Poly-Scientic 1213 North Main Street Blacksburg, VA 24060 (828)837-5115 Attn: Barbara Smith x231 fax (828)837-0846 Brushless DC motors McMaster Carr BOX 440 NEW BRUNSWICK, NJ 08903-0440 (908)329-3200 Mechanical Parts QUALITY COMPOSITES INC 8385 S ALLEN STREET No.140 SANDY, UT 84070 (801)565-8003 FAX: (801)565-8225 Carbon Fiber Tubes Sava Industries, Inc. 4 N. Corporate Drive, P.O. Box 30 Riverdale, NJ 07457-0030 (201) 835-0882 Steel Cable, crimps and crimp tools. Tristar Plastics Shrewsbury, MA 800.874.7827x3203(mike rudel) fax 508.845.1200

APPENDIX C. SUPPLIERS AND COSTS

53

Rulon bushings W.S. Deans 10875 Portal Drive Los Alamitos CA 90720 714-828-6494 Robin Wet noodle wire US Sensor Corp 1832 West Collins Ave. Orange, CA 92867 714-639-1000 Terry ext 104 Thermal sensors for mounting amps to heat sinks Tra-Con 45 Wiggins Ave. Bedford, MA 01730 1.800.872.2261 781.275.6363 Glue for glueing thermal sensors to amp heat sink. Tower Hobbies PO Box 9078 Champaign, IL 61824-9078 1-800-637-6050 Batteries and charger

C.1

Costs

This section contains the costs associated with the robot M2. Costs and quantities given do not include spares. The general lab policy is to order 50 percent spares on everything. I think that anyone else developing a similar robot should do the same. See Table C.1 for details.

APPENDIX C. SUPPLIERS AND COSTS

54

Component Actuator, Amp Custom Computer CFRP frame Machining Vestibular Misc Parts PCBs Load Cells TOTALS

QTY 12 1 8 1 1 200 10 8

Each US $ 1500 10,000 300 30,000 10,000 20 1,000 450

Total US $ 18,000 10,000 2500 30,000 10,000 4000 10,000 3600 88,100

Table C.1: Basic robot budget. See Robinson et al.((17)) for more detail on the actuator. This budget does not include prototyping or development costs. Part quantities and individual costs are not necessarily meant to imply identical parts but rather to give an average price for all units of a certain type.