Object Shape Characterisation using a Haptic Probe

Abstract

One of the fundamental challenges in robotics is object detection in unstructured environments where objects’ properties are not known a priori. In these environments sensing is susceptible to error. Conventional sensors, for example sonar sensors, video cameras and infrared sensors, have limitations that make them inadequate for successful completion of outdoor tasks where vision is partially or totally impaired. For instance, in harsh atmospheric conditions, the poor image quality that video cameras suffer and the low spatial resolution sonar sensors exhibit make them unsuitable for object detection. Therefore, we propose an alternative approach for object characterisation in unstructured environments using active touch. A simulated anthropomorphic haptic probe system with 3 DOF and a force sensor was used in this approach. The system measures and analyses forces at contact points, and data useful for object characterisation were realised. Two shapes, a sphere and a cuboid, were modelled and used in the object detection simulation and satisfactory results were obtained.