High-Speed Vision and Force Feedback for Motion-Controlled Industrial Manipulators

Abstract: Over the last decades, both force sensors and cameras have emerged as useful sensors for different applications in robotics. This thesis considers a number of dynamic visual tracking and control problems, as well as the integration of these techniques with contact force control. Different topics ranging from basic theory to system implementation and applications are treated. A new interface developed for external sensor control is presented, designed by making non-intrusive extensions to a standard industrial robot control system. The structure of these extensions are presented, the system properties are modeled and experimentally verified, and results from force-controlled stub grinding and deburring experiments are presented. A novel system for force-controlled drilling using a standard industrial robot is also demonstrated. The solution is based on the use of force feedback to control the contact forces and the sliding motions of the pressure foot, which would otherwise occur during the drilling phase. Basic methods for feature-based tracking and servoing are presented, together with an extension for constrained motion estimation based on a dual quaternion pose parametrization. A method for multi-camera real-time rigid body tracking with time constraints is also presented, based on an optimal selection of the measured features. The developed tracking methods are used as the basis for two different approaches to vision/force control, which are illustrated in experiments. Intensity-based techniques for tracking and vision-based control are also developed. A dynamic visual tracking technique based directly on the image intensity measurements is presented, together with new stability-based methods suitable for dynamic tracking and feedback problems. The stability-based methods outperform the previous methods in many situations, as shown in simulations and experiments.