A Hybrid DWA-MPC Framework for Coordinated Path Planning and Collision Avoidance in Articulated Steering Vehicles

Abstract

This paper presents an autonomous collision avoidance method that integrates path planning and control for articulated steering vehicles (ASVs) operating in underground tunnel environments. The confined nature of tunnel spaces, combined with the complex structure of ASVs, increases the risk of collisions due to path-tracking inaccuracies. To address these challenges, we propose a DWA-based obstacle avoidance algorithm specifically tailored for ASVs. The method incorporates a confidence ellipse, derived from the time-varying distribution of tracking errors, into the DWA evaluation function to effectively assess collision risk. Furthermore, the execution accuracy of DWA is improved by integrating a kinematic-based Model Predictive Control. The proposed approach is validated through simulations and field tests, with results demonstrating significant enhancements in collision avoidance and path-tracking accuracy in confined spaces compared to conventional DWA methods.