Sunteți pe pagina 1din 6

2015 IEEE Bombay Section Symposium (IBSS)

A Novel Automatic Low Cost Cutting Machine-cum-


3D Printer Using an Image Processing Based Control
Sudhir Rao Rupanagudi, Varsha G. Bhat, Vaishnav Vaddi Naga Padma Prasuna
Ram Savarni K.R., Sumukha Bharadwaj Atria Institute of Technology,
WorldServe Education, Bengaluru, India. Bengaluru,
sudhir@worldserve.in India.

Abstract— Over the past 5 years, there has been a tremendous manoeuvred the machine. Since this was purely based on
rise in designing various prototypes for a 3D printer. While most human control, there was no actual guarantee of any form in
of them concentrate on accuracy and speed, the cost has always refinement of measurement. As years progressed, Numerical
been a concerning factor. In this paper, we present a novel cost Control(NC) based systems came into light. These were first
effective setup which could be used for both the X and Y axis invented by John T. Parsons [11] in the year 1940. The early
motion, which is required for either a 3D printer or a cutting systems used punch cards where different cards corresponded
machine which is widely used in any industry. The paper also to different jobs. With the advent of computers in the early
concentrates on creating a CAD tool independent interface with 1960’s the focus shifted to providing inputs to these cutting
inputs to the setup being given purely based on images drawn.
systems via computer programs and this methodology of
This simplifies the process with the user not needing to have any
computer knowledge at all. All image processing algorithms
providing control was christened “Computer Numerical
were designed and developed using Simulink – a part of Control” (CNC) [12]. Early CNC machines involved
MATLAB 2011b and controls to the setup were given using an providing measurements and input in the form of a
Arduino Uno (ATmega328) Rev. 3 board, in turn creating a programming language. This proved tedious to the user, since
perfect embedded system setup. defining a complicated shape with the help of a software
language is a near impossible task. A major improvement to
Keywords—3D Printer; Arduino Uno; CAD; CNC; Cutting this methodology was the invention of the Computer Aided
tool; DNC; Embedded system; Image Processing; Laser Cutting; Design (CAD) software in which the user can draw the shape,
Microcontroller depth and size of the job just by using a mouse or a stylus as
an input [13]. Recent CAD tools even allow the user to draw
I. INTRODUCTION in 3D making it a very easy task even for the most
complicated objects [14]. With progress in time, there are
With the heralding of the “Make in India” movement by newer variants of CNC machines known as DNC – Direct or
the Honourable Prime Minister of India, the past two years has Distributed Numerical Controlled machines in the market
displayed a huge spurt in research, development and [15]. In these systems, the instructions to the machine are sent
innovation [1]. Several organizations and individuals have block by block from a separate computer either wirelessly or
made much advancement in science and technology in various via an IP address.
fields such as agriculture [2], healthcare, education and also
the retail sector. From smart shopping carts [3] to cloud Though all the aforementioned methodologies (both CNC
computing based automatic pest detection systems [4], a lot of and DNC) prove to simplify work for the end user, in a small
these discoveries help mankind in several different ways. scale industry perspective it proves to be an economically
Similar types of automations can also be seen in the industry infeasible solution. The major reason for this is the initial cost
sector as well, with a lot of scope given to reduced production of the CAD software and also of the cutting equipment [16].
and maintenance costs along with reduced labour requirement. In developing nations importing such tools also involve huge
expenditure. Also, training the users in the software can be an
Given any type of industry, one of the major tasks arduous task especially if complex designs are involved and if
performed on a daily basis is cutting and welding [5]. the software is not available in the regional language of the
Automation of the same can be seen in the automobile and user.
aeronautic industry [6]. Cutting equipment can also be seen in
use in the lumber industry for cutting wood [7] and also in the This paper presents a novel methodology to overcome the
food industry to cut and process fish [8]. Several problems mentioned above, by enabling the user to input the
methodologies to cut material exist, varying from using a shape to be cut in the form of an image sans the CAD tool.
simple diamond edge knife [9] to a complicated LASER Not much research has been carried out in this regard whilst
system [10]. In all the above mentioned cases, controlling the those who have, convert the input image to a G-Code once
cutting equipment is one of the major tasks and requires a again requiring a CAD tool. The existing methodologies to
great deal of precision and accuracy. perform the same have been elaborated in Section II. In order
to make the setup more affordable, a simple homemade
In the early years cutting was performed using a remote embedded system setup which could be used to cut small
control based approach wherein the user manually

978-1-4673-9542-7/15/$31.00 ©2015 IEEE


shapes, has been explained in Section III. Since the setup
already involves both X-axis and Y-axis motion, the same
setup could be used for a 3D printer. Modifications to be made
have also been included in the same section. Section IV
focuses on the software section – the image processing
algorithm to identify the shape to be cut along with the control
system to give instructions to the setup. The results obtained
have been shown in Section V and the future of this research
has been detailed in Section VI.
Fig. 1. Setup diagram of the prototype used in our experiments

II. EXISTING METHODOLOGY As shown in Fig. 1, the setup mainly consists of two major
As mentioned apriori, there is very little literature available sections: one - the input unit, and the second - the actual X-Y
wherein a camera has been utilized to input data to a CNC moving mechanical base plate. The input unit comprises of a
machine. In most cases, the camera is used to detect any plain white paper, with dimensions 21 cm x 14.85 cm, on
manufacturing defects after the process of cutting is complete. which the shape has to be drawn and also a camera to capture
For instance in [17] and [18], camera’s are deployed to the image of the shape drawn.. The camera employed here is a
observe defects on wood and micro electrodes respectively. 20 megapixel i-Ball Robo K20 camera [25], with image
resolution of 5500 x 3640 pixels. The camera is fixed to a
Image processing has also been used in [19], where the stand at a height of 30 cm, such that it can focus the complete
authors define a way to identify the wear and tear involved in sketch area.
the cutting tool of a machine, using a camera. At the same
time, no provision has been made to use the camera to give The main components of the mechanical unit are a
inputs to control the same. conveyer belt and three DC motors. One of the DC motors is
utilized to move the conveyor belt in the x-axis direction. Two
In [20] and [21], the authors do discuss a methodology to separate DC motors are used along with wheels to assist in the
use a camera to identify the grooves of a key and pass that as Y-Axis motion of the system.
an input to a CNC system which duplicates keys. The method
describes an image processing flow to identify various aspects The conveyer belt is built on a wooden frame made up of
of a key, convert the same to G-Codes and feed the same to timber, whose dimensions are 35.2 cm x 25.2 cm x 12.5 cm. In
the CNC. The algorithm though accurate and precise will order to offer smooth rotation of the belt, five hollow
work only for duplicating keys and not any other simple aluminium rods of length 25 cm and diameter 1.5 cm are
shapes. Also a requirement of conversion to G-Code once placed parallel to each other in the frame. The major challenge
again persists and no mention of the setup to manufacture the of the complete setup was to select the best suited material for
key has been described. the belt. The material that has to be chosen should not be too
tight since it might damage the motor.
In [22], the authors introduce a new type of CNC which
involves the use of CAD oriented CNC (CADoCNC). This The first material of choice was a 100% cotton cloth.
involves utilization of image processing algorithms on a Unfortunately, since the cotton cloth was too soft, it did not
scanned image of the job at hand. Processes such as edge provide sufficient friction between the belt and the rods for its
thinning, noise removal and color conversion are involved and motion. So in order to improvise friction, a rough material had
the instructions are further converted to a dxf file which is to be considered. The second material of choice was then a
then fed to the CNC machine. More elaboration of the same Gunny bag, with dimensions same as that of the cotton cloth
can be seen in [23] and [24]. Similar to [21] there is once and due to its roughness, provided the right amount of friction.
again a dependency on the CAD tool and the results were The disadvantage of using the Gunny material was the lack of
tested using an emulator instead of using an original setup. elasticity. This created a huge force on the metal pipes causing
them to bend inwards. Finally, after a lot of experimentation,
In order to overcome these shortcomings, a cost effective an elastic band of 1 inch width was chosen and this worked
setup has been described in the next section and the algorithm perfectly well due to its property of being both elastic and
to identify the shape to be cut has been elaborated in detail in rough.
Section IV.
In order to rotate the conveyer belt, a DC motor is
III. SETUP connected to one of the aluminium rods. The input voltage
provided to the motor is 12V and provides a speed of 100 rpm.
As mentioned in Section I, the main novelty of the paper, In order to provide a regulated and constant voltage supply to
lies in the prototype which can be utilized for either cutting the DC motor and also to power on-off the motor when
small shapes or can also be utilized as a base for printing 3D required, an L293D Motor Control Board is utilized. The
objects. Another major feature of our setup is the way in controls to this motor board are provided from an Arduino
which inputs can be given. Instead of a traditional CAD tool, Uno Board which in turn takes decisions based on the image
the prototype receives input in the form of images drawn on a processing performed apriori in Simulink.
simple sheet of paper. The complete block diagram of our
prototype can be seen in Fig. 1. For the y-direction motion, the complete wooden frame
with conveyer belt is placed on two heavy duty wheels
controlled by two DC motors. To balance the frame, castor mammoth task for the processor to complete analysis of the
wheels, also known as omni-directional wheels are utilized. picture in the shortest time possible. In our application, since
These DC motors are also controlled by another set of L293D color is actually a redundant information (as the drawing
motor boards along with the same Arduino Uno. created is anyways in black and white), the image hence can
be converted to a suitable color model to reduce the total
The base plate on which the shape has to be carved or on number of bytes per pixel.
which the model is to be printed, has to be placed on the
conveyer belt and its maximum weight can be approximately
Obtain image of the shape to be
30 grams, for free motion of the conveyor belt. The dimension
printed or cut
of the cutting plate can be 21.8 cm x 16 cm. A stepper motor
setup along with a rack and pinion arrangement can also be
utilized for z-axis motion inturn assisting in printing 3 Convert it to intensity plane
dimensionally as well.
The complete setup mentioned above has been shown in
Fig. 2. Yes Is θ1 ≤ pxl No
≤ θ 2?

Output pixel Output pixel is


is made white made black

Removal of lines to retain dots

Image Inversion

Fig. 2. Hardware model of our setup utilized for cutting or 3D printing Count number of Vertices
The image processing algorithm utilized to control the
Arduino Uno and in turn the motors, has been described in Find distance between any two
detail in the next section. vertices

IV. OUR METHODOLOGY


Send shape and distance to control
In this section we describe the algorithm utilized to unit
identify the shapes to be printed or cut. The steps involved to
perform the same can be seen in Fig. 3 and have been further Fig. 3. Flowchart of the algorithm utilized to identify the size and shape of
elaborated below. Since this is an initial prototype, simple the figure drawn
shapes such as a square, rectangle and a triangle have been
taken into consideration. The algorithm could later on be In this algorithm, the picture taken is converted to a
extended to more complex shapes. grayscale (Y) color model defined by the ITU-R
BT.601standard [26] and this is guarded by (1) [27].
A. Capturing the image to be cut/printed
The initial step in the whole process is to first capture the Y' = (0.299 × R) + (0.587 × G) + (0.114 × B) (1)
image of the shape that is to be cut or printed. As mentioned in
the previous section, the shape under consideration is drawn The main advantage of conversion, as mentioned above, is
on a white A4 size sheet and captured using a web camera the reduction in number of bytes required to represent each
placed above. In order to ease the extraction of the image in pixel. Instead of requiring 3 bytes per pixel to store the image,
later stages, the vertices of the shape drawn are thickened in a mere 1 byte pixel would now suffice.
comparison with the rest of the image i.e. the lines.
C. Segmentation
B. Color Conversion The next important step of the algorithm is to segment or
It is a well known fact, that image processing on the whole, segregate the object of interest from the background. The
consumes a lot of processing time due to vast number of pixels segmentation process is defined by (2). The entire image pixel
of the picture under consideration. On an average, a 640x480 area is scanned and if any pixel lies between a predefined
color image is made of around 921,600 pixels, making it a range θ1 and θ2, then that particular pixel is made white else it
is made black. With this step, the memory requirement for
each pixel further drops since only 1 bit is required per pixel Scan image for white pixel and save
after this step. (R,C) in the array

Safety factor δ added to R


1 Y(x, y) > θ1 && < θ2
S(x, y) =
0 otherwise (2) Jump  columns to right

Once completed, the whole white area of the paper is made


Continue scanning row
white and the shape under consideration black. Since only the
vertices are required noise removal is a must and is explained
next.
Yes Hit white No
D. Noise Removal pixel?
The main objects of interest are the vertices and not the
lines joining them. In order to eliminate the same, a Save coordinate R+ 
morphological operation called dilation is adopted [28]. The (R+δ,c1)
main feature of dilation is to eliminate any black pixel which
is surrounded by atleast a single white pixel. Since the lines
are of small width, and are surrounded by white pixels, the R+
dilation process removes all lines of the shape which connect
the vertices.

E. Restoring integrity of the vertices and image inversion


Since dilation performed in the previous step eliminated No Is R =
most of the black pixels, a few of the pixels belonging to the
maxrow?
vertices also disappear. In order to restore the same, a
morphological operation opposite to that of dilation is Yes
performed. This is known as erosion [29]. Upon performing
this step the vertex dots regain back most of their lost pixels. Find the difference between column
coordinates on same row in the array
Since the further processing stages require the dots to be
white and the background black, a simple image inversion Fig. 4. Flowchart to calculate the distance between any two vertices of a
operation is performed to achieve the same. shape

F. Determining the shape


A simple counting algorithm such as the contour tracing
algorithm [30] is deployed which in turn assists in counting
the total number of vertices. Since simple shapes are taken
into consideration the total number of vertices would in turn
directly correspond to a particular shape. For instance, if the
number of vertices is 3, it would naturally correspond to a
shape that of a triangle.
Fig. 5. Pixel profiling of vertices for the distance calculation algorithm
G. Determining the distance
The last and most important step in the algorithm is to The image of the shape obtained after erosion is now
determine the distance or size of the shape. This can be easily scanned from the top right hand corner until a white pixel is
performed by measuring the distance between any two encountered. The co-ordinates of the first white pixel obtained
vertices on the same row of an image. The difficulty lies in the is labelled as (R,C) and is shown in Fig. 5. Since this pixel is
fact that each vertex is made of more than one pixel and this not the exact centroid of the first vertex, a safety factor δ is
could lead to multiple distance calculations between pixels of added to the row number. This was chosen based on several
different vertices. In order to overcome this problem a new experiments conducted over a large data set. Since the camera
algorithm to measure distance is proposed and shown in Fig. height is fixed from the paper, the value of δ does not change
4. over time. These new co-ordinates (R+δ,C) are now stored in
an array. The safety factor also helps in case the user has not
For more clarity and to understand the concept better, two drawn the vertices in a perfect straight line.
vertices belonging to a square have been zoomed and labelled
in Fig. 5.
The next step is to obtain the co-ordinates of the second
vertex. This can be done by scanning the picture, starting off
from (R+δ,C) and searching for the next white pixel. But it
can be clearly seen from the image, that the next white pixel
would belong to this particular vertex itself. In order to
overcome this, another safety factor α is added to the column
number (R+δ, C+ α), as shown in Fig. 5, so that the first (a) (b) (c)
vertex is completely surpassed. Scanning of the image
continues from (R+δ, C+α) until a subsequent white pixel is
again encountered. The co-ordinates of the same (R+δ,C1) are
once again stored in the aforementioned array. Once the end of
the row is reached, in order to take care not to recount the
same vertex, another safety factor of β is added to the row
(d) (e)
number (R+δ) and the process continues till the last row is
reached. Fig. 6. An image of a square given as inout to the 3D printer and various
stages of processing of the same. Fig. 6a - A picture of a square drawn on
In this way, the co-ordinates of all vertices present in the paper of different shades - as seen by the camera. Fig. 6b - Conversion of the
photograph taken are stored in an array. Once this completes, square into grayscale. Fig. 6c - Resultant image obtained after dilation with
the array is scanned and the co-ordinate values of any two sides of the square obliterated. Fig. 6d - Image obtained after erosion –
column co-ordinates belonging to a same row are extracted obtaining back the lost pixels of the vertices. Fig. 6e - Output of image
and subtracted to obtain the distance. This is shown in inversion
equation 3.
The prototype was tested for various types of shapes
ranging from a square, rectangle and a triangle. Out of all
d = |(C-C1)| (3) attempts 100% accuracy was obtained. Initial stages of the
experiments conducted can be seen in Fig. 7 wherein the very
Other algorithms such as Euclidean distance [32] could first square drawn using the prototype can be seen.
also be used depending on the application.
Once the shape and distance is obtained, this information
is further passed on to the control unit located on a
Microcontroller. The microcontroller is preloaded with
instructions on how the base plate must move in both X-axis
and Y-axis for different shapes. The number of times and
duration each motion must complete is purely decided by the
size of the shape using a Look up table based approach.
The next section deals with the results obtained at each
stage of this algorithm.
Fig. 7. The first square drawn with the prototype using a suspended pen
V. RESULTS
The complete image processing algorithm was designed
and developed using the Simulink tool bundled with Matlab VI. CONCLUSION AND FUTURE SCOPE
2011b. The controls required to cut/print different shapes were In this paper a novel methodology to design and develop a
loaded onto an Arduino Uno - ATmega328 Rev.3 low cost cutting machine-cum-3D printer has been presented.
microcontoller [31]. Commands from Simulink were sent By utilizing only an image as an input and with the image
serially through the COM port to the Arduino based on the processing algorithm later on offloaded on a dedicated
shape and distance obtained. The outputs obtained at every processor (ASIC), the prototype would be totally CAD
stage of the algorithm have been shown below. independent and also would not require a CNC setup. The user
could give the specifications of the design just by drawing it
Fig. 6a shows the picture of the square drawn on a sheet
on a piece of paper and the design would then be completed.
with varying shades, as visible to the camera. The image
This requires the user to have neither prior computer
obtained after converting to gray scale has been shown in Fig.
knowledge nor that of any CAD tool, ideal for rural areas of
6b. The output of the dilation process can be seen in Fig. 6c
developing nations.
with the lines of the square clearly obliterated and the vertices
diminished in size. It must be noted that variation in the The major aspect of the work carried out is the cost
background shade of the paper does not affect the algorithm in effectiveness as well. On survey over the internet, the cheapest
anyways. Fig.6d shows the reconstruction of the vertices using available 3D printer on the net is just over $240 [33]. Upon
the erosion operation and Fig. 6e - the result obtained after calculation and addition of all extra features including a rack
image inversion. and pinion arrangement with a motor for the z-axis motion, the
setup described in this paper costs a mere $65. This would
prove to be a huge advantage especially for small scale
industries and also rural areas wherein a reduced initial setup [16] Shtereva, K.; Genchev, I., "Teaching integrated circuits design using
cost could give a boost to their economical growth. As a future LASI," Advanced Topics in Electrical Engineering (ATEE), 2013 8th
International Symposium on , vol., no., pp.1,4, 23-25 May 2013
scope more complicated designs and shapes can be
[17] Conners, R.W.; et.al "Identifying and Locating Surface Defects in
experimented upon. A much larger prototype would also be of Wood: Part of an Automated Lumber Processing System," Pattern
great use. Deployment in the small scale industry would be the Analysis and Machine Intelligence, IEEE Transactions on , vol.PAMI-5,
next major step. no.6, pp.573,583, Nov. 1983
[18] Guo, Rui; Zhao, Wansheng; Li, Gang, "Development of a CNC micro-
EDM machine," Technology and Innovation Conference, 2006. ITIC
REFERENCES 2006. International , vol., no., pp.705,709, 6-7 Nov. 2006
[1] Make in India, “MAKE IN INDIA” [Online]. Available: [19] Kalaichelvi, T.; et.al, "Automated intelligent machine vision system for
http://www.makeinindia.com [Accessed May 15, 2015]. monitoring the image," Signal and Image Processing (ICSIP), 2010
[2] Huddar, S.R.; Gowri, S.; Keerthana, K.; Vasanthi, S.; Rupanagudi, S.R., International Conference on , vol., no., pp.462,466, 15-17 Dec. 2010
"Novel algorithm for segmentation and automatic identification of pests [20] Itthisupornrat, V.; Kiatsunthorn, S., "Duplicate a blank key using image
on plants using image processing," Computing Communication & processing," SICE-ICASE, 2006. International Joint Conference , vol.,
Networking Technologies (ICCCNT), 2012 Third International no., pp.5715,5718, Oct. 2006
Conference on , vol., no., pp.1,5, 26-28 July 2012
[21] Cheewanantakul, M.; Kiatsunthorn, S., "Duplicate a key using image
[3] Rupanagudi, S.R.; Jabeen, F.; Savarni K R, V.R.; Adinarayana, S.; processing," SICE Annual Conference, 2008 , vol., no., pp.40,44, 20-22
Bharadwaj, V.K.; Karishma, R.; Bhat, V.G., "A novel video processing Aug. 2008
based cost effective smart trolley system for supermarkets using
[22] Tan Liu; Tao Yu; Wenbin Wang, "INC: a new type of computer
FPGA," Communication, Information & Computing Technology
numerical control," Computer Supported Cooperative Work in Design,
(ICCICT), 2015 International Conference on , vol., no., pp.1,6, 15-17
Jan. 2015 2005. Proceedings of the Ninth International Conference on , vol.2, no.,
pp.787,792 Vol. 2, 24-26 May 2005
[4] Rupanagudi, S.R.; Ranjani, B.S.; Nagaraj, P.; Bhat, V.G.; Thippeswamy,
[23] Tao Yu; et.al, "Application of Image Processing Technology in
G., "A novel cloud computing based smart farming system for early
INC," Intelligent Control and Automation, 2006. WCICA 2006. The
detection of borer insects in tomatoes," Communication, Information &
Sixth World Congress on , vol.2, no., pp.6203,6207, 0-0 0
Computing Technology (ICCICT), 2015 International Conference on ,
vol., no., pp.1,6, 15-17 Jan. 2015 [24] Tan Liu; Tao Yu; Wenbin Wang; Shuzhen Yang, "Research on CAD
[5] Mazumder, J., "Advanced laser processing of metals," Advanced technology oriented to CNC," Mechatronics and Automation, 2005
IEEE International Conference , vol.4, no., pp.1750,1755 Vol. 4, 2005
Applications of Lasers in Materials Processing/Broadband Optical
Networks/Smart Pixels/Optical MEMs and Their Applications. [25] iBall, “iBall Robo K20.0 Webcamera” [Online]. Available
IEEE/LEOS 1996 Summer Topical Meetings: , vol., no., pp.23,25, 5-9 http://www.iball.co.in/Product/ROBO-K20/833. [Accessed May. 15,
Aug. 1996 2015].
[6] Jun Cai; Zongxing Ding; Yi Zhang; Min Liu, "Trajectory planning and [26] Radiocommunication Sector of ITU, Recommendation ITU-R BT.601-7,
simulation for intersecting line cutting of the industry robot," Intelligent Studio encoding parameters of digital television for standard 4:3 and
Control and Automation (WCICA), 2014 11th World Congress on, vol., wide-screen 16:9 aspect ratios, March 2011
no., pp.63,68, June 29 2014-July 4 2014 [27] Dhruva, N.; Rupanagudi, S.R.; Sachin, S.K.; Sthuthi, B.; Pavithra, R.;
[7] Sang-Mook Lee; et.al, "Automated Grading, Upgrading, and Cuttings Raghavendra, "Novel segmentation algorithm for hand gesture
Prediction of surfaced dry hardwood lumber," Image and Signal recognition," Automation, Computing, Communication, Control and
Processing and Analysis, 2009. ISPA 2009. Proceedings of 6th Compressed Sensing (iMac4s), 2013 International Multi-Conference on ,
International Symposium on , vol., no., pp.371,376, 16-18 Sept. 2009 vol., no., pp.383,388, 22-23 March 2013
[8] de Silva, C.W.; Wickramarachchi, N., "An innovative machine for [28] Prashanth, C.R.; Sagar, T.; Bhat, N.; Naveen, D.; Rupanagudi, S.R.;
automated cutting of fish," Mechatronics, IEEE/ASME Transactions on , Kumar, R.A., "Obstacle detection & elimination of shadows for an
vol.2, no.2, pp.86,98, June 1997 image processing based automated vehicle," Advances in Computing,
Communications and Informatics (ICACCI), 2013 International
[9] Yajing Shen; Nakajima, M.; Kojima, S.; Homma, M.; Fukuda, T., "Nano
Conference on , vol., no., pp.367,372, 22-25 Aug. 2013
knife fabrication and calibration for single cell cutting inside
environmental SEM," Micro-NanoMechatronics and Human Science [29] Rupanagudi, S.R.; Vikas, N.S.; Bharadwaj, V.C.; Sowmya, K.S.; Manju;
(MHS), 2010 International Symposium on , vol., no., pp.316,320, 7-10 Dhruva, N., "Novel methodology for blink recognition using video
Nov. 2010 oculography for communicating," Advances in Electrical Engineering
(ICAEE), 2014 International Conference on , vol., no., pp.1,6, 9-11 Jan.
[10] Bruhm, Hartmut; et.al, "A Motion Control Strategy for Robots in Laser
2014
Material Processing and other High Speed Applications," Robotics
(ISR), 2010 41st International Symposium on and 2010 6th German [30] Hedberg, H.; Kristensen, F.; Owall, V., "Implementation of a Labeling
Conference on Robotics (ROBOTIK) , vol., no., pp.1,6, 7-9 June 2010 Algorithm based on Contour Tracing with Feature Extraction," Circuits
[11] Ward, J.E., "Numerical Control of Machine Tools," McGraw Hill and Systems, 2007. ISCAS 2007. IEEE International Symposium on ,
vol., no., pp.1101,1104, 27-30 May 2007
Yearbook of Science and Technology, McGraw-Hill, New York, 1968,
pp. 58-65. [31] Arduino Uno, “Arduino Uno” [Online]. Available
[12] Acramatic CNC. Improved man/machine communications," Production http://www.arduino.cc/en/Main/ArduinoBoardUno. [Accessed May. 15,
Engineer , vol.54, no.6, pp.367,369, June 1975 2015].
[32] S. Rupanagudi, V. Rupanagudi, M. Kamuf, and V. Öwall, “Reducing
[13] Dlugatch, Irving, "The applicability of computer-aided design as a
computational complexity of branch metric calculations in a trellis
system engineering tool," Proceedings of the IEEE , vol.55, no.11,
pp.1940,1945, Nov. 1967 decoder,” presented at the International Symposium on Wireless
Personal Multimedia Communications (WPMC), Saariselkä, Finland,
[14] Lachenmaier, J.F.; Lasi, H.; Kemper, H.-G., "A Concept for Extracting 2008.
and Sharing Technical Data from Digital Product Models for Subsequent
[33] 3D printer prices, “Price compare – 3D printers” [Online]. Available:
Processing," System Sciences (HICSS), 2015 48th Hawaii International
Conference on , vol., no., pp.987,996, 5-8 Jan. 2015 http://www.3ders.org/pricecompare/3dprinters/. [Accessed May. 15,
2015]
[15] Jiang Guodong, "Direct Numerical Control Techniques applications in
local area network," Electronic and Mechanical Engineering and
Information Technology (EMEIT), 2011 International Conference on ,
vol.7, no., pp.3388,3390, 12-14 Aug. 2011

S-ar putea să vă placă și