Vision-based collision avoidance algorithms in a collaborative robotic cell for human-robot interaction

Human-robot cooperation is increasingly relevant in industrial contexts, especially for tasks like the assembly and disassembly of complex systems, where human flexibility complements robotic precision. To enable safe and efficient collaboration, this work presents a novel collaborative robotic cell integrating a mobile base, an articulated manipulator, and a network of depth cameras for real-time human tracking. A key contribution is the spatial synchronization framework, which employs vision systems and fiducial markers to align human and robot coordinate systems. Control algorithms based on artificial potential fields ensure dynamic trajectory adaptation and collision avoidance. The proposed architecture, tested in simulation and on a physical prototype, demonstrates effective coordination in shared workspaces and shows promise for broader applications in flexible manufacturing.