A Human Perception System for Robot Collaborative Tasks

Today’s humans need to collaborate with robots more and more frequently by sharing a common task space. The cooperation often translates into a physical collaboration, i.e., a context in which a physical contact occurs between the two agents. The physical Human-Robot Interaction (pHRI) delves into aspects and mechanisms of such interaction and it is subject to a huge interest from the scientific community. Robotics research is currently engaged in the design and development of robotic devices that are able to cooperate with humans. In industry, for example, cobots have to actively collaborate with humans while ensuring safety and ergonomic conditions for the workers. Although today’s robots are equipped with all sorts of sophisticated sensors (e.g., cameras for in-depth vision, inertial sensors, force and tactile sensors, powerful actuators), the onboard technologies do not always guarantee the task success when the collaborative task with a human demands a physical contact. If the task is not previously scheduled in the robot control loop, the quality of the interaction decreases because of the unpredictability of the human movements [1]. To ensure the ergonomic quality of the interaction, the robots have to perceive the human partners to mimic the implicit mutual awereness of two collaborative humans [2] [3]. From the robot perspective, the perception is equivalent to i) know the position of the human, ii) detect the motion of the human (e.g, velocities, accelerations), iii) quantify the exchanged contact forces and the human joint torques for the ergonomy control of the task [4]. In general, the perception constitutes of a package of instantaneous human data to be exploitable by the control loop of the robot to adjust the interaction strategy, Fig. 1. This abstract highlights a perception system for human kinematics and dynamics (i.e., joint torques, internal forces exchanged across joints, external forces acting on links). We introduce a novel framework endowing robots with the ability to perceive humans while fulfilling a common cooperative task. The framework has been developed as a part of the European Union H2020 funded project An.Dy [5] which focuses on strengthening the European robotics market and moving research results out of the lab and into the marketplace.