Force, acceleration and vision sensor fusion for contour following tasks with an industrial robot

In robotic contour following tasks such as sewing or cutting an industrial robot guides a tool along a contour of a workpiece. Teaching the task manually is very time consuming and the process would be unable to react to uncertainties or changes of the environment. Thus sensor systems are applied to reduce human intervention, making the robot more autonomous. In this paper we present an integrated approach for multisensor contour following. A look-ahead vision sensor guides the tool along a contour of a workpiece while force feedback control maintains a desired contact force. Couplings that arise between these two controllers are estimated and compensated through online measurements of contact forces between robot and environment. The identification of the environment combines force and acceleration sensors in an observer-based control scheme. The system is validated by experiments that involve contour following on compliant objects