ο€  Advances in Technology Innovation, vol. 2, no. 2, 2017, pp. 40 - 45 40 Position Control and Novel Application of SCARA Robot with Vision System Hsiang-Chen Hsu1,2,*, Li-Ming Chu3, Zong-Kun Wang2 and Shu-Chi Tsao2 1 Department of Industrial management, I-Shou University, Kaohsiung, Taiwan. 2 Department of Mechanical and Automation Engineering, I-Shou University, Kaohsiung, Taiwan. 3 Department of Mechanical Engineering, Southern Taiwan University of Science and Technology, Tainan, Taiwan. Received 02 February 2016; received in revised form 27 April 2016 accepted 02 May 2016 Abstract In this paper, a SCARA robot arm with vi- sion system has been developed to improve the accuracy of pick-and-place the surface mount device (SMD) on PCB during surface mount process. Position of the SCARA robot can be controlled by using coordinate auto- compensa- tion technique. Robotic movement and position control are auto-calculated based on forward and inverse kinemat ics with enhanced the intelligent image vision system. The determined x-y position and rotation angle can then be applied to the de- sired pick & p lace location for the SCARA robot. A series of experiments has been conducted to improve the accuracy of pick-and-place SMDs on PCB. Keywords : SCARA, pick-and-place, forward and inverse kinematics, vision system 1. Introduction Industrial heavy duty manipulator such as Selective Compliance Assembly Robot Arm (SCARA) is an automatic device which is capable to carry and move components /devices/ parts in the manufacturing process. The first SCARA robot was invented by Professor Hiro- shi Makino from University o f Yamanashi, Japan in 1978 [1]. Fig.1 demonstrates the basic structure includes kinetic 4-axis and 4-DOF, translation on X, Y, Z and rotation about vertical Z-axis. Two-link compliant arms with rotated wrist behave somewhat like the human arm that joints allow the arm to move vertically and horizontally in a limited space. SCARA was specially designed for precision devices Fig. 1 The first SCARA robot by Hiroshi Makino [1] assembly, especially place a pinned component in a hole. A typical SCARA robot is a stationary robot arm including base, elbow, vertical extension and tool roll and comprising both rotary and prismat ic joints. SCARA robots may vary in size and shape but they all are consistent in a unique 4-axis motion [2]. W ith this distinguished feature, SCARA particularly fits the pick-and-place sur- face mount devices on PCB (printing circuit board) and move-to-take delicate silicon wafers or glass panels on magazine. SCARA robot was introduced in SEIKO watch assembling lines in 1981 and since then industrial SCARA robot has been widely used in electronic, semiconductor, automobile, electronic, plastic, food and pharmaceutical factories [3] all over the world. SCARA robot is the principle use of robotics field and most of the domain of robotics field is on industry and academia. The control system of industrial SCARA robot is a highly non-linear, * Corresponding author, Email: hchsu@isu.edu.tw Advances in Technology Innovation, vol. 2, no. 2, 2017, pp. 40 - 45 41 Copyright Β© TAETI strong coupling and time-vary ing systems [4]. Kinemat ics modeling is one of the key tech- nologies to verify the model. Many researchers and engineers [5-6] have presented the manipu- lating ability of robotic mechanis ms in pos i- tioning and orienting end-effectors and propose a measure of manipulability. The motion tra- jectory of a robot arm is calcu lated using the geometric analysis. The PID control techniques [7] have been proposed to solve the nonlinearity issue with acceptable results to control the movement of robot arms. The motion trajectory of a robot arms is calcu lated using the geometric analysis with Matlab software. The performance of industrial heavy duty robots working in unstructured environments can be improved using visual perception and learning techniques [7-10]. The object recogni- tion is accomplished using an artificial neural network (ANN) arch itecture. A novel technique used in the assembly lines integrates computer vision to capture the shape of the objects, online grasp determination based on that shape, and image-based control for grasp execution. Visual servoing system [8] consists of high-speed im- age processing, kinematics, dynamics, control theory, and real-time computing to control the position and orientation of a robot with respect to an object. Furthermore, the color of objects could also be considered and included in the image-base system. The developed image pro- cessing and workpiece recognition algorithm is based on LabVIEW Vision Development Mod- ule [9]. Later, like two eyes on human, two cameras are developed to detect the robotic arm movement in 3-D space. In this system, the robotic arm is controlled and moved, and after mathematic calcu lations the precise position of the motors is calculated to reach the designated position [10]. The automation in surface mount precision assembly lines often consists of SCARA robots equipped with grippers, image v ision system and linked by motorized conveyances. In this system, the combination of high performance motion control with integrated vision guidance and conveyor tracking are demonstrated. The robotic arms are used for the pick-and-place of the SMT components. The placement system on printed circuit board (PCB) is mainly influenced by the surface mount device (SMD) robots and the production environment in assembly line. Temperature on the reflow process is often above 250 O C which easily distorts the tray. The deformed tray would result in misalignment of the placement system. The yield rate is conse- quently descended and the cost would be enor- mous. The purpose of this paper is to improve the accuracy of the placement system on PCB dur- ing surface mount process . Coordinate au- to-compensation technique on deformed t ray is developed to control the position of SCARA robots equipped with image vision system. Ro- botic movement and position control are calcu- lated based on forward and inverse kinematics . 2. Theoretical Development Fig. 2 illustrates the coordinate system of random point located on PCB (x, y) with respect to robotic arm. Fig. 2 Random point (x, y) on PCB 2.1. Forward Kinematics Location of end effector can be determined by the length of robotic arm and rotation angle of each axis 𝑋 = 𝐿1π‘π‘œπ‘ πœƒ1 + 𝐿2π‘π‘œπ‘ (πœƒ1 + πœƒ2 ) (1) π‘Œ = 𝐿1π‘ π‘–π‘›πœƒ1 + 𝐿2𝑠𝑖𝑛(πœƒ1 + πœƒ2 ) (2) where L1, L2 are robotic arm length, and 1, 2 are rotation angle of each axis Advances in Technology Innovation, vol. 2, no. 2, 2017, pp. 40 - 45 42 Copyright Β© TAETI 2.2. Inverse Kinematics Rotation angle of each axis can be determined by the coordinate system of end point . 2 2 2 2 1 1 2 2 1 2 cos 2 X Y L L L L  ο€­  ο€­ ο€­ ο€½ (3)     1 2 21 1 1 2 2 Β  tan tan L sinY X L L cos      ο€­ ο€­ ο€½ ο€­   οƒΆ ο€½ ο€­   οƒ·   οƒΈ  οƒΈ (4) 2.3. Compensation for Image and Distance Coordinate Fig. 3 demonstrates the captured image of pixel with respect to distance (1 pixel equals to 0.01 mm). The origin o f d istance coordinate is located in the center of image coordinate and then pixel (320, 240) would be the same as dis- tance (3.2mm, 2.4mm). Fig. 3 Image/distance coordinate system 2.4. Compensation for deformed PCB Tray Coordinate Fig. 4 presents the compensation for de- formed PCB tray coordinate. Assuming the PCB tray offset 1 mm due to thermal induced warp- age, the coordinate of random point in image capture system is determined. 𝑋2 = 𝑃𝐢𝐡 𝑋 + 𝑃𝑖π‘₯𝑒𝑙 𝑋 (5) π‘Œ2 = 𝑃𝐢𝐡 π‘Œ + 𝑃𝑖π‘₯𝑒𝑙 π‘Œ (6) where X1=28.2, Y1=52.4 are system default parameters. After mathemat ical calculation, the placement coordinates are 𝑋 = 𝑋2 βˆ’ 𝑋1 (7) π‘Œ = π‘Œ2 βˆ’ π‘Œ1 (8) Fig. 4 Schematic illustration of PCB coordinate system 2.5. Image Processing Grayscale dig ital image is a range of shades of gray without apparent color. The reason for differentiating gray images is that only specify a single intensity value for each pixel, i.e. less informat ion for each pixel. In order to reduce the complexity of post-processing grayscale image scheme is applied to capture the characteristic image. The Hough transform is a technique which can be used to isolate features of a particular shape, such as line, circle, ellipse, etc., within an image. Based on Hough transform, circular detection on LabVIEW Vision Assistant is ap- plied to determine the p recise position on de- formed PCB. 3. Structure of Improved Placement System 3.1. Hardware Description A Toshiba SCARA robot (#SR-424SHP) with control card (#PISO-PS400) shown in Fig.5 is employed in this study. Fig. 6 illustrates SMD components placed on PCB with tray (carrier). The placement system on trial run conveyer is presented in Fig. 7. Advances in Technology Innovation, vol. 2, no. 2, 2017, pp. 40 - 45 43 Copyright Β© TAETI Fig. 5 Toshiba SCARA robot (#SR-424SHP) Fig. 6 SMD components on PCB with tray (carrier) Fig. 7 Placement system on trial run conveyor The overall placement system also consists of trial run conveyor for PCB tray, SMD co m- ponents feeder, buzzer, 3 cylinders (baffle, push and charge-in). 2 cameras (coordinate au- to-compensation and poka-yoke), 3 sensors with Arduino UNO and a lighting system (top and broadside). All SMD components are held in specific trays which are loaded in upstream vibratory parts feeding stations. 3.2. Software Development The control software used in this study is LabVIEW 2012 and LabVIEW Vision Assistant 2012. 4. Results and Discussion Coordinate compensation system for SMD placement and starved feeding (queliao) au- to-detection system have been developed in this paper. In the first, anchor point on PCB was shot by camera and the offset amount on PCB was then calculated by compensation of the image and distance coordinate. The precise location on PCB for components placement was then de- termined. For queliao and those parts did not place in the desired position, the developed system can also sound a buzzer signal on the control annunciator panel and warn the upstream feeding stations to load SMD components. Fig. 8 Flow chart of coordinate compensation system for SMD placement Fig. 8 and Fig. 9 present the flow chart for coordinate compensation system and queliao auto-detection system, respectively. Fig. 10 demonstrates the accuracy of SMD components pick-and-place improved by the developed al- gorithm. The accuracy on auto assembly (99.73%) has been dramatically improved after teaching mode (78.66%). Start Capture SMD Image Grayscale Digital Image Hough Transform Calculation Pixel Position for Point (X, Y) Stop Advances in Technology Innovation, vol. 2, no. 2, 2017, pp. 40 - 45 44 Copyright Β© TAETI Fig. 9 Flow chart of queliao auto-detectionsystem Fig. 10 Accuracy of SMD components pick-and-place 5. Conclusions In this paper, an improved pick-and-place SMD on PCB system has been accomplished by using SCARA robot and machine v ision. The result has shown that 97.33% yield rate was achieved and more than 2% of erro r could be eliminated by improving the hardware of system. Besides top lighting camera, anchor point on the skewed PCB tray can be calibrated and revised by a broadside webcam. Remote operation using wireless network is feasible. Acknowledgement The authors would like to express their ap- preciation to Ministry of Science and Technol- ogy, Taiwan, ROC, for financial supports under project No. MOST103-2221-E-214-018 and MOST104-2221-E-214-051. Appreciation is also extended to Oriental Semiconductor Electronic, Ltd. for carrying out all the experiments. References [1] SCARA, The ROBOT Hall of Fame, Power by Carnegie Mellon, http://www.robothalloffame.org/inductees/ 06inductees/scara.html [2] A. Burisch, J. Wrege, A. Raatz, J. Hesselbach, and R. Degen, β€œPARVUS – miniaturised ro- bot for improved flexibility in micro produc- tion,” Assembly Automation, vol. 27, no. 1, 1980. [3] S. K. Dwivedy and P. Eberhard, β€œDynamic analysis of flexible manipulators, a literature review,” Journal of Mechanism and Machine Theory, vol. 41, no. 7, pp. 749-777, 2006. [4] A. Visioli and G. Legnani, β€œOn the trajectory tracking control of industrial SCARA robot manipulators,” IEEE Transactions on Industrial Electronics, vol. 49, no. 1, pp. 224-232, 2002. [5] G. S. Huang, C. K. Tung, H. C. Lin, and S. H. Hsiao, β€œInverse kinematics analysis trajectory planning for a robot arm,” Proceedings of 8th Asian Control Conference (ASCC 2011), Kaohsiung, Taiwan, pp. 965-970, May 2011. [6] J. Fang and W. Li, β€œFour degrees of freedom SCARA robot kinematics modelling and simulation analysis,” International Journal of Computer, Consumer and Control, vol. 2, no. 4, pp. 20-27, 2013. [7] F. Escobar, S. DΓ­az, C. GutiΓ©rrez, Y. Ledeneva, C. HernΓ‘ndez, D. RodrΓ­guez, and R. Lemus, β€œSimulation of control of a SCARA robot ac- tuated by pneumatic artificial muscles using RNAPM,” Journal of Applied Research and Technology, vol. 12, no. 5, pp. 939-946, 2014. [8] S. H. Han, W. H. See, J. Lee, M. H. Lee , and H. Hashimoto, β€œImage-based visual servoing control of a SCARA type dual-arm robot,” IEEE International Symposium on Industrial Electronics, Cholula, Puebla, Mexico, vol. 2, pp. 517-522, Dec. 2000. [9] H. Zhu, J. Xu, D. He, K. Xing, and Z. Chen, β€œDesign and implementation of the moving workpiece sorting system based on LabVIEW,” 26th Chinese Control and Decision Conference Start Image Captured Image mask Advanced Morphology Stop Part Analysis http://ieeexplore.ieee.org/xpl/tocresult.jsp?isnumber=21159 Advances in Technology Innovation, vol. 2, no. 2, 2017, pp. 40 - 45 45 Copyright Β© TAETI (2014 CCDC), Changsha, China, pp. 5034-5038, May 2014. [10] R. Szabo and A. Gontean, β€œRobotic arm control with stereo vision made in LabWind ows/CVI,” 38th International Conference on Telecommunications and Signal Processing (TSP), Prague, Czech, pp. 1-5, July 2015.