Using the Microsoft Kinect to Model the Environment of an Anthropomimetic Robot

Abstract

The control of compliantly actuated anthropomimetic robots with complex and multiarticular joints, such as those developed within the ECCEROBOT project, is extremely challenging. We are approaching the problem by using a physics engine to run a highly detailed simulation of such a robot’s structure and dynamic behaviour, and then search-ing for sequences of motor activations that will achieve par-ticular goals. This requires the simulated robot to be situ-ated accurately in a physics-based representation of its en-vironment which includes the object with which it is to in-teract. In this paper we present our environmental sensing and modelling scheme which uses data from a single head-mounted Kinect sensor to provide and locate the environ-mental model, and to identify and locate the target object accurately in the presence of significant motion blur.