Multisensor Robotic Contour Following on Deformable Objects

Heiko Koch, Alexander König, Karl Kleinmann, Alexandra Weigl-Seitz, and Jozef Suchý

Keywords

Multisensor application, Computer Vision, Contour Following, Industrial Robot

Abstract

Sensory perception of the environment in robotics reduces the amount of necessary human intervention within manufacturing processes. Autonomous contour following tasks, such as sewing, grinding or applying adhesives require a vision system to guide the robot along a desired path on a workpiece. Moreover, by force feedback control a desired contact force between workpiece and tool can be maintained. Using multiple sensors in one control scheme, especially with rather compliant environments causes couplings between force and vision control. In this paper we derive a decoupling algorithm that compensates for deformations of the environment within the vision sensor data. System delays and data filtering are taken into account. Necessary environmental parameters are gained by online-identification. We present experimental results of constrained robotic contour following, using a hybrid control structure, to validate this approach.

Important Links:



Go Back