Farrokh Janabi-Sharifi
Ryerson University, Canada
Title: Visual Servo Control of Robotic Systems: Applications and Challenges
Biography
Biography: Farrokh Janabi-Sharifi
Abstract
Conventional robotic systems operate on open loop kinematic chain for positioning tasks and hence they are not robust to operate in unstructured environments. Uncertainties in target object’s position, unplanned compensation for object’s motion, bending of the links due to the load and accelerations, and joint slippages are examples of the elements frequently encountered in many industrial operations. Significant effort and expenses are therefore extended to enable robots’ operation in uncertain environments. Reprogramming and calibration of robots for recurrent or new tasks also add to the cost of robots’ operations. In this talk, I will review the fundamentals of direct end-point control of robots using vision sensor, i.e., visual servoing, and will show how this framework could overcome many of the aforementioned practical problems. The main structures and theoretical as well as practical issues in development and integration of visual servo control will be reviewed. Recent results in my laboratory including sensor fusion schemes, uncertainty modeling, and robust control for visual servoing will be shown. Next, I will show how integration of visual servoing would facilitate robots programming. A recently developed novel vision-based programming by demonstration approach (PbD) will also be discussed and its practical implications will be demonstrated.
