Feasibility study into the suitability of collaborative robots for various manufacturing operations
Koh, Yu Wei Jonathan
Date of Issue2016
School of Mechanical and Aerospace Engineering
A*Star Advanced Remanufacturing and Technology Centre
In recent times, as technology advances and becomes increasingly sophisticated, robots are no longer constrained to simultaneous localization and mapping for control. A new alternative of programming robots was introduced when Microsoft released the Kinect which allowed the integration of an economical and compact camera that is able to capture 3-D coordinates in real time to manipulate them. The integration of the Microsoft Kinect for Windows v2 camera sensor with the Universal Robots UR10 robot arm is carried out in this study. The main objectives are to establish the accuracy of the UR10 robot arm in mimicking the hand joints motion of a human being by using the Kinect for Windows v2 camera sensor, carry out repeatability tests to validate the consistency of the Kinect for Windows v2 camera sensor, and to verify the margins of error. Through the employment of Microsoft Visual Studio 2013 to write the main body in C++ language, the codes facilitated the combination of colour, depth, and joints of the Kinect sensor to work cooperatively. Python codes were written to integrate the C++ codes cohesively to connect the UR10 robot arm to the PC workstation for communication. Lastly, a UR script interfaced with the PC workstation to read and write data which resulted in the execution of the UR10 robot arm. The accuracy of the UR10 robot arm is evaluated to mimic the right hand placements of the user on a table top. Through the tracking of the right hand tip of the user by the Kinect for Windows v2 camera sensor, the UR10 robot arm is tasked to pinpoint the same exact locations. The results obtained compare both predicted and actual movements and showed similar trends in all three axes of the 3D space coordinates. The repeatability of the Kinect for Windows v2 camera sensor is evaluated and analysed to ensure the location points recorded by the Kinect are accurate every time through a series of points that are repeated five times each. The results obtained suggested that an elevated height above the work top table yields more precise recording of coordinates as compared to a non-elevated setup.
Final Year Project (FYP)
Nanyang Technological University