Efficient Local and Global Sensing for Human Robot Collaboration with Heavy-duty Robots

Human robot collaboration (HRC) with heavy-duty industrial robots is required in various production and recycling processes. They require optimal sensing methodology, which ensure safety from collision while allowing high robot velocity. Variety of local and global sensing approaches exist in industrial context. However, they are either only applicable for small robots, or limit the maximum robot velocity. This work proposes a novel integrated local and global sensing methodology for optimal collision detection and avoidance. The sensing methodologies is realized by complying to speed regulation from TS15066. It is realized by a) transforming robot model in the two sensing reference frames, b) estimating parallel shortest distance between the human and the robot in the sensing reference frames and finally by c) regulating the robot velocity based on relative human position. The global sensing ensures the robot deceleration as the human moves towards the robot. This allows using constant sized local search zones, which reduce false detections with close proximity worker at robot acceleration. The proposed system is validated using a heavy-duty industrial robot for a constant human presence. The methodology allows 11% more process efficiency compared to global only sensing system, while allowing 0% increase in the production space requirement, which makes it applicable to retrofit previously, installed robotic cells.