DRPA-MPPI: Dynamic Repulsive Potential Augmented MPPI for Reactive Navigation in Unstructured Environments

2025-03-26

AI Summary

To address trajectory failure and susceptibility to local minima when mobile robots encounter unexpected obstacles in unstructured environments, this paper proposes an MPPI-based reactive navigation framework enhanced with dynamic repulsive potentials. Our method introduces three key contributions: (1) a novel online local-minimum detection mechanism leveraging trajectory curvature and gradient variation of the distance field; (2) a target-directed and repulsion-optimized dual-mode adaptive cost function, eliminating the need for pre-training or manual parameter tuning; and (3) real-time dynamic distance field integration to construct adaptive repulsive potentials, significantly improving environmental adaptability. Evaluated in dense obstacle simulations, the proposed framework achieves a 37% increase in navigation success rate, a 62% reduction in collision rate, and a 28% decrease in computational overhead—demonstrating substantial improvements in both safety and real-time performance.

Technology Category

• Planning, Routing, and Scheduling: Replanning and Plan Repair
• Intelligent Robots: Motion and Path Planning
• Search and Optimization: Learning to Search

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

Reactive mobile robot navigation in unstructured environments is challenging when robots encounter unexpected obstacles that invalidate previously planned trajectories. Model predictive path integral control (MPPI) enables reactive planning, but still suffers from limited prediction horizons that lead to local minima traps near obstacles. Current solutions rely on heuristic cost design or scenario-specific pre-training, which often limits their adaptability to new environments. We introduce dynamic repulsive potential augmented MPPI (DRPA-MPPI), which dynamically detects potential entrapments on the predicted trajectories. Upon detecting local minima, DRPA-MPPI automatically switches between standard goal-oriented optimization and a modified cost function that generates repulsive forces away from local minima. Comprehensive testing in simulated obstacle-rich environments confirms DRPA-MPPI's superior navigation performance and safety compared to conventional methods with less computational burden.

Problem

• Reactive navigation in unstructured environments with obstacles
• Overcoming local minima traps in MPPI control
• Adapting to new environments without heuristic costs

Innovation

• Dynamic repulsive potential for obstacle avoidance
• Automatic switching between cost functions
• Enhanced navigation performance with less computation