OPEN PLATFORM FOR REAL-TIME ROBOTIC VISUAL SERVOING

This paper presents an open platform to design, simulate and test robotic image-based visual servoing systems using off-the-shelf commercial cameras and standard personal computers. First a set of simulation files for Matlab is in-troduced and then a real-time toolkit for visual servoing is presented. The latter makes extensive use of separate threads and multiprocess computing to support the image and video interfaces as well as the network backbone link between each computer in the system. These toolkits are used to design an eye-in-hand image-based visual servo-ing system for real-time tracking in the TQ MA2000 robot which exhibits high backlash and low resolution. Simula-tion and real-time results are documented in the paper. An overview of the functions included in the real-time toolkit is also provide