Human-robot interaction for industrial robots

• Main Author: Su, Hansel Zhenwei
• Other Authors: Pham Quang Cuong
• Format: Final Year Project
• Language: English
• Published: 2018
• Subjects: DRNTU::Engineering::Mechanical engineering::Robots
• Online Access: http://hdl.handle.net/10356/75537
• Institution: Nanyang Technological University

In this report, a programming method for an industrial robot using a motion capture system is designed and implemented. This report integrates several existing work to create the interface to create a pipeline to ‘teach’ the robot. The software is built with ROS, a middleware for building robotic applications and the motion planning algorithms are implemented in OpenRAVE, a software architecture for implementing motion planning algorithms. The programming method in this report involves ‘drawing’ a path above a surface, and have the robot execute it afterwards by moving the end effector to trace the path. First, using the Optitrack cameras to track a rigid body with retroflective markers, we ‘draw’ a path. Then the recorded path is downsampled, using a modified version of Ramer-Douglas-Peucker algorithm to remove redundant or duplicate waypoints, reducing the computation time for the algorithms that follow. The modified algorithm ensures that the waypoints are not too far apart after reducing the sample size. Following that, a graph search method adapted from RoboTSP is used to search for the optimal IK solution for each waypoint in the recorded path. The path is then time parameterized, to satisfy the velocity and acceleration constraints of the robot. The recorded paths are then verified on a Denso VS060 6-DoF industrial manipulator model in OpenRAVE by means of a maximum deviation metric, which compares the final geometric path with the original path.