Machines (Aug 2023)

Dynamic Human–Robot Collision Risk Based on Octree Representation

  • Nikolaos Anatoliotakis,
  • Giorgos Paraskevopoulos,
  • George Michalakis,
  • Isidoros Michalellis,
  • Evangelia I. Zacharaki,
  • Panagiotis Koustoumpardis,
  • Konstantinos Moustakas

DOI
https://doi.org/10.3390/machines11080793
Journal volume & issue
Vol. 11, no. 8
p. 793

Abstract

Read online

The automation of manufacturing applications where humans and robots operate in a shared environment imposes new challenges for presenting the operator’s safety and robot’s efficiency. Common solutions relying on isolating the robots’ workspace from human access during their operation are not applicable for HRI. This paper presents an extended reality-based method to enhance human cognitive awareness of the potential risk due to dynamic robot behavior towards safe human–robot collaborative manufacturing operations. A dynamic and state-aware occupancy probability map indicating the forthcoming risk of human–robot accidental collision in the 3D workspace of the robot is introduced. It is determined using octrees and is rendered in a virtual or augmented environment using Unity 3D. A combined framework allows the generation of both static zones (taking into consideration the entire configuration space of the robot) and dynamic zones (generated in real time by fetching the occupancy data corresponding to the robot’s current configuration), which can be utilized for short-term collision risk prediction. This method is then applied in a virtual environment of the workspace of an industrial robotic arm, and we also include the necessary technical adjustments for the method to be applied in an AR setting.

Keywords