An Adaptive Distance-Velocity Weighted Force Field for Real Time Obstacle Avoidance in Unmanned Ground Vehicles

Abstract

Unmanned Ground Vehicle navigation remains a critical challenge in dynamic and unstructured environments. This paper proposes an Adaptive Beta-Weighted Force Field method for real-time obstacle avoidance, in which the repulsive force coefficient adapts continuously based on obstacle distance, velocity, and type classification. Unlike conventional fixedweight force field approaches, the proposed method modulates the repulsive gain proportionally to proximity and mobility of each detected obstacle, enabling the vehicle to respond conservatively at range while reacting at close encounters. Three configurations are evaluated through simulation: a fixed variedweight baseline, a fixed uniform-weight control, and the proposed adaptive-weight method, all tested under identical obstacle environments combining static and dynamic obstacles with an A-star global planner. Results demonstrate that the adaptive method achieves the fastest navigation completion while maintaining stable and smooth force behavior throughout the trajectory. The avoidance force variability is substantially reduced compared to both baselines, indicating smoother motion generation. Path efficiency remains comparable across all methods, confirming that adaptive weight modulation improves navigation speed and force stability without sacrificing safety or path quality. The adaptive method achieves a navigation time of 17.20 seconds, a mean avoidance force of 2.848 N, and a force standard deviation of 4.361 N, representing a 22% reduction in travel time and 74% reduction in force variability compared to the fixed-weight baseline, while maintaining a path following efficiency of 96.6%.