Safety in Admittance Control using Reference Trajectory Shaping

A team of researchers has introduced a novel control framework designed to make robots inherently safer for working alongside humans. The paper, 'Safety in Admittance Control using Reference Trajectory Shaping,' presents a switched model reference admittance controller. This system dynamically shapes a robot's planned movement (reference trajectory) and adjusts its compliance (admittance parameters) in real-time based on both task requirements and strict safety boundaries. A key innovation is the explicit integration of 'invariance control,' a mathematical guarantee that ensures the robot's motion never leaves a pre-defined safe region, even in the presence of external disturbances.

The researchers provided a rigorous stability analysis using a common quadratic Lyapunov function, proving the system's tracking error converges to zero. They demonstrated the controller's effectiveness through simulations on a complex two-link manipulator, showing superior performance compared to existing methods. Crucially, they also implemented it in real-time on a physical single-link manipulator, validating its practical feasibility. This real-world test highlights the controller's dual ability to maintain compliant, natural-feeling interaction while enforcing hard safety constraints, a vital step for advanced human-robot collaboration in manufacturing, healthcare, and assistive robotics.