Future service robots are expected to achieve high-quality task performance for everyday household chores. Some of the most frequent tasks in this domain are related to wiping of surfaces, such as vacuuming the floor, sweeping dust, or cleaning windows. However, the performance for these tasks is not directly observable as small dirt particles, dust, and residual water are hardly perceivable by means of computer vision. In this work we propose to utilize haptic perception paired with a qualitative effect representation to reason about the task performance of robotic wiping motions despite poor visual information. In particular, we relate the desired contact force to the measured end-effector force in order to simulate the effect of previously executed wiping motions. This way we are not just able to distinguish good from bad contact situations, but also replan recovery motions w.r.t. the effect-space to accomplish the commanded cleaning task subsequently. We evaluate our approach in a set of experiments with the robot Rollin’ Justin.
Daniel Leidner and Michael Beetz “Inferring the Effects of Wiping Motions based on Haptic Perception”, in Proc. of the International Conference on Humanoid Robots (ICHR), Mexico, November 2016.
A draft version of the research paper is located at: http://elib.dlr.de/107678/
More information on Rollin’ Justin: http://rmc.dlr.de/justin