Dynamic obstacle avoidance for robotic arms via steering force field modeling

Dynamic obstacle avoidance for robotic arms remains challenging in confined workspaces when obstacles are irregular and moving. This paper presents a steering force field modeling approach embedded in Dynamic Movement Primitives for task space trajectory generation with obstacle avoidance. The method introduces generalized distance and generalized angle, together with affine obstacle transformation and virtual circle envelopes, enabling unified handling of irregular and freely moving obstacles. Compared with baseline methods, the proposed approach produces smoother obstacle-avoiding trajectories, lower free-space loss, and favorable real-time computational efficiency. A Lyapunov-based analysis shows that the coupling design preserves goal convergence while avoiding spurious local minima and sustained oscillations. Furthermore, under explicit assumptions, a higher-order control barrier function analysis establishes collision avoidance guarantees. Numerical simulations and experiments on a Franka Emika Panda robot executing planar end effector motions validate the method in both static and dynamic scenarios, demonstrating its effectiveness for generating reliable obstacle-avoiding trajectories for robotic arm motion planning in constrained dynamic environments.