SIGMO: Sigmoid Gated Interaction-Aware Motion Planning

This paper introduces SIGMO, a motion planning framework that uses polynomial sigmoid functions to analytically estimate yield scores of surrounding agents based on physics-grounded interaction features. The method integrates these functions into a receding-horizon planner for interaction-aware planning with arbitrary trajectory predictions. Validation is performed through interaction-intensive simulations and hardware experiments, demonstrating that SIGMO balances proactive and defensive behavior while remaining computationally lightweight.