SafeDMPs: Integrating Formal Safety with DMPs for Adaptive HRI
Pranav Tiwari, Soumyodipta Nath, Ravi Prakash
AI summary
Problem
Robotic systems require both robust trajectory generation and formal safety guarantees, but current approaches force a trade-off: efficient methods lack safety proofs, while formally safe methods rely on computationally expensive online optimization unsuitable for high-frequency control.
Approach
The framework uses Dynamic Movement Primitives for nominal motion planning and integrates a closed-form, optimization-free control law from Spatio-Temporal Tubes to enforce time-varying safety envelopes and adaptively steer around obstacles.
Key results
- Orders-of-magnitude faster execution than NODE-CLF-CBF baselines
- Zero collisions with formal safety guarantees across static and dynamic obstacle scenarios
- Smooth, bounded perturbation recovery with faster convergence than baselines
- Real-time hardware validation on a 7-DOF Franka robot under human interference
Why it matters
Provides a computationally efficient, provably safe control framework for deploying robust robots in dynamic, human-centric environments.
Abstract
Robots operating in human-centric environments must be both robust to disturbances and provably safe from collisions. Achieving these properties simultaneously and effi- ciently remains a central challenge. While Dynamic Movement Primitives (DMPs) offer inherent stability and generalization from single demonstrations, they lack formal safety guarantees. Conversely, formal methods like Control Barrier Functions (CBFs) provide provable safety but often rely on computa- tionally expensive, real-time optimization, hindering their use in high-frequency control. This paper introduces SafeDMPs, a novel framework that resolves this trade-off. We integrate the closed-form efficiency and dynamic robustness of DMPs with a provably safe, non-optimization-based control law derived from Spatio-Temporal Tubes (STTs). This synergy allows us to generate motions that are not only robust to perturbations and adaptable to new goals, but also guaranteed to avoid static and dynamic obstacles. Our approach achieves a closed-form solution for a problem that traditionally requires online opti- mization. Experimental results on a 7-DOF robot manipulator demonstrate that SafeDMPs is orders of magnitude faster and more accurate than optimization-based baselines, making it an ideal solution for real-time, safe, and collaborative robotics.