Improving Accuracy of Local Maps with Active Haptic Sensing

Abstract

Walking robots are supposed to be ubiquitous urban reconnaissance machines. To make them fully operational, accurate local maps of the environment are required. Here the local map is the local representation of the environment (obstacles), for example stairs. There are several perception systems which provide data used to create such representations. In this paper experiments with three of them are described: a Laser Range Finder, a Time of Flight Camera, and a Stereo Camera. The data provided by the above mentioned sensors is not fully reliable. Thus there is a need to augment the perception system with active haptic sensing, understood here as sensing the environment with the legs of the robot. At the beginning of the article the results of the experiments with the sensors mentioned above are described. Then the active sensing concept is presented in detail. Next, an example of using the active perception system is demonstrated. At the end concluding remarks are given, followed by future work plans.