Integration of Vision based Robot Manipulation using ROS for Assistive Applications

Providing vision to robots has been an effective application in the field of robotics which enhanced automation. There have been several algorithms developed in recent times for object recognition and YOLO v3 which is one of the best has shown very good results in real-time object recognition. Consequently, the same has been chosen for this research. This research aims at developing a vision based autonomous manipulation system for assistive applications such as service robots, rescue robots, wheelchairs, etc. A detailed system integration between vision and robotic arm using the Robot Operating System (ROS) framework has been presented. This paper also explains the method of coordinate transformation between camera and manipulator reference systems and the results are discussed.