RRT$^\eta$: Sampling-based Motion Planning and Control from STL Specifications using Arithmetic-Geometric Mean Robustness
The framework solves multi-constraint scenarios 3x faster than traditional STL-based planners by smoothing optimization landscapes.
Researchers Ahmad Ahmad, Shuo Liu, Roberto Tron, and Calin Belta developed RRT^η, a sampling-based motion planning framework. It replaces traditional min-max robustness with an Arithmetic-Geometric Mean (AGM) measure to evaluate Signal Temporal Logic (STL) specifications. This creates smoother optimization, enabling more efficient exploration. The system was validated on three robots—a point robot, unicycle, and 7-DOF arm—synthesizing control sequences that satisfy complex spatiotemporal constraints with high probabilistic completeness.
Why It Matters
Enables robots to autonomously execute complex, safety-critical tasks defined in human-readable logic, advancing industrial and service robotics.