Proprioception for collision detection
Document typeMaster thesis
Rights accessOpen Access
Nowadays a lot of research is done in the eld of human-robot interaction to allow robots to work in human environments. Haptic perception is crucial for human and robot safety since it allows to detect contacts. It is also important for surface exploration be able to detect contact between a robot and its environment. This thesis presents a probabilistic representation of contacts in order to complement or in some cases replace computer vision systems. Torque measurements were used to determine probable collisions (i.e. positions and forces) on robot links. The suggested algorithm were tested in simulation and afterwards experiments were done with the robot arm Kuka LWR.