Sunteți pe pagina 1din 10

Autonomously Controlled Vehicle

MM2EM2

Author : Kalp Patel Student ID : 4173646 Design Team : Bangs-mobile Submission Date: 25/11/2013

UNIVERSITY OF NOTTINGHAM | University Park, Nottingham, NG7 2RD, UK

Contents
1. Introduction .......................................................................................................... 0 1.1 Abstract ............................................................................................................. 0 1.2 Statements of requirements for the prototype robot .......................................... 1 2. Concept Evaluation .............................................................................................. 1 2.1 Design 1 ............................................................................................................ 2 2.2 Design 2 ............................................................................................................ 2 2.3 Design 3 ............................................................................................................ 3 3. Concept Selection ................................................................................................ 4 3.1 Discussion ......................................................................................................... 4 3.2 Embodied prototype design............................................................................... 5 4. Controller Operation ............................................................................................. 6 4.1 Bang-bang (on-off) controller ............................................................................ 6 4.2 PID (Proportional-Integral-Derivative) controller................................................ 6 4.3 LabView code.................................................................................................... 7 5. Conclusions ......................................................................................................... 7 6. Bibliography ......................................................................................................... 8

1. Introduction
1.1 Abstract
Warehouse robots are increasingly providing a safer, reliable and efficient solutions to companies that handle large amounts of warehouse stock. In a design team of three, the aim was to design and build a small scale prototype stock-handling robot for a company that are interested in upgrading their warehouse with autonomously controlled vehicles (ACVs). The warehouse floor will be marked with lines which can

be picked up using a light sensor and processed by the ACV to follow that path. An overhead lighting system will regulate ACVs motion and safety. The prototype robot demonstrated the capabilities of this approach by meeting a set of statements of requirements shown below. The prototype was made using the Lego Mindstorms NXT kit was and programmed using the National Instruments LabView control software. Drop off area This swerve tests both the right and left ACV steering

Pick up area

Figure 1 The two courses under which the ACV prototype was tested on

1.2 Statements of requirements for the prototype robot


Be completely autonomous and require no human interference whilst operating. Navigate itself around using a highly reflective coloured line (2.5 cm thick) which will be laid down on the floor of significance different in colour. This requirement was later amended to a black line (2.5 cm thick) on white floor. Be able to securely transport a small load; a tennis ball (weight between 56.0 g 59.4 g and 6.7 cm diameter)[3] from one end to the other end as shown in fig. 1. Be efficient and complete as many circuits as possible in a given time frame. Be able to stay on track; follow straights and turns by changing direction. Stop instantly and take no action when a red light is shone on it from above. Keep operating when the red light is absent. Only use NI LabView control software to program the control system.

2. Concept Evaluation
A total of three designs were considered after a group concept generation activity. One design per every member of the design team. All three concepts had to implement: A secure place for the tennis ball. A light sensor positioned close to the ground facing downwards. A colour sensor facing upwards.

2.1 Design 1
This tracked vehicle design was considered due to its compact robust framework. The load is spread out on the rubber tank treads due to the large contact area and thus it can carry heavier loads. It provides good traction at a cost of lower speeds. At high speeds; slippage of the tank treads occur. At low speeds; the slippage of the tank treads occur arbitrarily. This uncontrollable slippage was mainly due to the Lego Mindstroms NXT kit track treads not fitting properly around drive wheels. Rubber bands were fitted around the drive wheels to tackle this problem, which made the slippage occur less frequently but did not completely resolve the issue. The steering control was excellent as the movement is based on two separately driven motors placed on either side. It can thus change its direction by varying the relative rate of rotation of its wheels and hence does not require an additional motion or control program.

Figure 2 Tank-like tracked vehicle front and side view sketches

Advantages Very robust framework Carry heavier loads Differential steering [2] Simple control system

Disadvantages Aesthetically unappealing Low operating speeds Prolonged use places strain on the motors and the mechanics of the tank treads. Uncontrollable slippage due to the tank tread

2.2 Design 2
This design tackled some of the disadvantages of design 1 but raised various other problems. The typical Ackermann steering control made the design very aesthetically pleasing for which a complex control system would have to be deployed to control the extra steering motor at the front. This Ackermann steering control would not allow sharp turns unlike the differential steering in which the right motor can turn in the clockwise direction and the left in the anticlockwise direction. The four large wheels divide the loads equally and help maintain high operating speeds. The extra added parts due to the Ackermann steering make this design less robust and prone to regular maintenance which would add extra cost per ACV on the company, making this design expensive and unreliable. Placing the light sensor behind the front wheels means there will be a slight delay in motion which may lead to a minor 2

systematic lag in the system. This design has back a wheel drive gear ratio of 3:1 which makes the ACV extremely fast and effective but reduces safety and may be prone to dropping stock when braking heavily and turning sharply.

Figure 3 Car design with Ackermann steering front and side view sketches

Advantages Aesthetically appealing Very high operating speeds Can carry heavy loads

Disadvantages Ackermann steering Complex control system Unreliable Expensive Lengthy

2.3 Design 3
A balance of designs 1 and 2 can be found in this design. The differential steering control allows it to turn sharply which uses less floor space per turn which is effective when considering a warehouse full of ACVs (a low turning radius [7]). This low turning radius allows it to even spin on one spot which can be desirable as the ACVs are meant to turn around 360 (make a U-turn) and go back once the stock is transported to collect more stock. An ideal steering control would be to reverse the motor controls instead to avoid full rotations which wastes time and reduces efficiency. The smaller rear wheels had full 360 of freedom (in one plane) and allowed the robot turn in any direction as well as spread the loads over four wheels, allowing it carry relatively heavy loads. The position of the colour sensor is implemented in such a way that it adds to the cage-like framework that securely stores the tennis ball. Overall this design uses more parts than design 1 but considerably less than design 2.

Figure 4 Design 3 side view by Khalid Ali

Advantages Aesthetically appealing High operating speeds Differential steering Simple control system Robust

Disadvantages Weaker rear wheels

3. Concept Selection
3.1 Discussion
Design 3 was the final chosen concept due to its many advantages listed above and only one disadvantage. A decision matrix was not needed as it was clear it will prove to be the ideal choice which was demonstrated to be true during the testing and tutorial evaluation. The differential steering with wheels without the rubber tank treads would solve all of the issues faced with using rubber tank treads at a cost of not being able to carry very heavy loads, but it would have better control and a simpler control system. Better control was deemed more important as it was one of the statements of requirements and the prototype only had to carry one tennis ball which weighs between 56.0 g 59.4 g which is relatively light. The design was also more aesthetically pleasing than design 1. The use of gears to make the ACV faster like in design 2 was not needed as it was required to operate safety and reliably. A faster ACV would be more efficient but may drop the tennis ball and will brake stop in a very uncontrollable manner. The tennis ball is placed on top of the Lego Mindstorms NXT brick which is the strongest part of all the designs and caged securely to prevent it from falling or slipping off. The smaller rear wheels were weaker than the larger front wheels which did not divide the load evenly relative to the front and back of the ACV. The wheels had to be kept small to allow a small 4

turning radius by having full 360 of freedom (in one plane). Larger wheels at the back would restrict the 360 of freedom and make the ACV have a higher centre of mass which is not ideal for carrying loads and risks toppling over whilst operating.

Figure 5 Design 3 front and side view sketches by Kalp Patel

3.2 Embodied prototype design

Figure 6 Front-left view

Figure 7 Back-right view

As seen from figure 6, the front light sensor is placed right at the front of the ACV to provide it time to process the information and control the motors accordingly. The light sensor is placed on the left hand side at the front and colour sensor on the right hand side. This was to allow the gap needed for the wires that feed into the Lego Mindstorms NXT brick at the front. The light sensor being on the left also allows the ACV to follow the black line centrally as the light sensor reads the left hand edge of the black line. Also, as seen from figures 6 and 7, the motor wires are very long. This was because of the Lego Mindstorms NXT kit lacking shorter wires of the ideal length needed. The colour sensor was placed facing upwards as required to detect the overhead lighting system. Yellow Lego parts were used for the front sensor holder as it represents caution of tripping, falling and striking hazards. The Occupational Safety and Health Administration (OSHA)[4] requires that workplace hazards be marked to alert employees to dangers that exist at a facility or worksite. Thus, the colour yellow will be ideal for the full scale ACVs, making them clearly 5

visible to the workers on the warehouse floor and minimizing the risks of injury by the ACVs.

Figure 8 Attachable tow-trailer

Figure 9 Fully loaded ACV

As discussed earlier on how the design 3 cannot carry as much load as design 1; a slight modification can be made to design 3 by attaching a small tow-trailer to the ACV in order can carry an extra tennis ball or just a tennis ball in the trailer to make the load spread out lengthwise in order eliminate the only one disadvantage of design 3.

4. Controller Operation
To control the ACV, two feedback controllers were proposed as explained below. Along with that a decision had to be made on whether to use: One light sensor or two light sensors. High sensitivity EOPD light sensor or low sensitivity standard light sensor.

4.1 Bang-bang (on-off) controller


This controller has two states it choses from; on and off. The control system code is simple in that if the ACVs light sensor following the left hand edge of the black line encounters white a surface beyond a certain threshold, (i.e a right bend) then accelerate the left motor by a set amount and vice versa. The issue with this approach is that robot will follow the black line in a zig-zag motion after encountering its first bend and is never actually straight, even when the line is straight. This zigzap motion can be damped a little by using two light sensors following the right and left hand edges of the black line and making the ACV more sensitive to change direction but the zig-zag nature of the motion will not disappear.

4.2 PID (Proportional-Integral-Derivative) controller


The PID controller is a superior controller compared to the bang-bang controller. An ideal PID system consisting of the three-term control; proportional, integral and derivative (PID), would give the best control, however using all three in conjunction will lead to high costs for the company. A simple PI controller or even just a P controller can provide the control system needed for the ACV. The proportional controller controls by taking in the required value; in this case the light sensor value and it compares it to the required light sensor value and calculates the amount of steering correction that is proportional to distance off of the left hand edge of the black line and smoothly brings it back to the edge to correct for this error. This provides smoother transitions compared to a bang-bang controller. A proportional gain constant is a multiple that is used to correct the amount of output error in response to the input error. The gain acts as a sensitivity control for the light sensor. 6

A higher proportional gain constant/sensitivity will lead to quicker servo motor response. A balanced gain constant has to be found to keep the ACV stable. An integral term can be added which uses the sum of the light sensor errors and provides an offset that should have been corrected previously. It acts as a controllers memory and fixes errors that keep persisting over time and effectively removes small errors.

4.3 LabView code


It was decided as a team, to use only one high sensitivity EOPD light sensor to keep the cost of the ACV down along with a simple control. A proportional controller was

3 1 the final chosen control method due to its simplistic effective control. Adding an additional integral controller was not needed as it would require a complex control system and it only makes the ACV somewhat more reliable.
Figure 10 ACV LabView proportional control system code

The EOPD light sensor placed outside of the while loop takes in a value when the ACV is placed on the edge of the black line and compares it to the EOPD light sensor inside the true case structure using a subtraction node. Then that value is multiplied by the proportional gain constant, 0.4, which is then passed to an addition and a subtraction node which are provide the power input of the right and left motors respectively. The 50 going into those nodes is the motor speed constant. A true/false case structure is made inside the while loop for the pause function of the ACV. This is done using the colour sensor, which is running continuously and provides a false Boolean value for when the colour sensor equals 5 i.e. red colour detected and a true Boolean value when it does not equal 5. For when it detects the overhead red light a simple brake all motors function is a put inside the false case structure which brings the ACV to a complete stop. The LabView code was loaded on to the Lego Mindstorms NXT brick to minimise the code execution time.

5. Conclusions
The aim of the project was achieved as all the statements of requirements for the prototype ACV were met. The capability of such small scale prototype ACVs were thoroughly evaluated and tested. During the LabView code testing some amendments had to be made to the code as the practice tracks were different to the final testing tracks and involved some very sharp turns. As a result the ideal proportional gain/sensitivity of the light sensors was put at 0.4 and the motor speed at only 50. During the final prototype demonstration the proportional gain/sensitive was raised to 1 and the motor speed was set to 70. Motor speed beyond 70 started 7

to make the ACV unstable. The ACV operated very effectively and achieved 13 stock transfers in 2 minutes. Future improvements can be made by making the ACV design more compact by using shorter wires and placing the sensors more securely. The LabView code can be improved by adding an integral and derivative controller terms along with an additional light sensor for maximum system control, stability and reliability

6. Bibliography
[1] University of Nottingham moodle page for MM2EM2 Electromechanical Systems 2 - http://moodle.nottingham.ac.uk/course/view.php?id=13692 [2] Differential steering control http://rossum.sourceforge.net/papers/DiffSteer/DiffSteer.html [3] International Tennis Federation for tennis ball information http://www.itftennis.com/home.aspx [4] Safety colour and coding labels for safe and healthful working conditions https://www.osha.gov/ [5] Comparison of a bang-bang controller to a proportional controller http://www.fourmilab.ch/hackdiet/www/subsection1_2_3_0_5.html [6] An existing line following robot http://www.nxtprograms.com/NXT2/line_follower/steps.html#Program [7] Turning radius http://en.wikipedia.org/wiki/Turning_radius [8] PID Control, Araki, Kyoto University, Japan http://www.eolss.net/ebooks/Sample%20Chapters/C18/E6-43-03-03.pdf

S-ar putea să vă placă și