A Stability and Terminal Constraints-based Obstacle Avoidance Framework for Autonomous Industrial Vehicles Under Complex Scenarios

Unmanned industrial vehicles are prone to dynamic instability during emergency obstacle avoidance. Variations in multi-dimensional stability over time pose significant challenges to both collision avoidance and stability control, resulting in hesitation and conservatism in vehicle decisions, similar to those of human drivers. This paper analyzes the avoidance distance within the context of multidimensional stability boundaries and proposes a framework that integrates high-level behavioral decision-making with low-level motion tracking. First, the pure braking distance and safety distance are calculated considering the vehicle's longitudinal and lateral stability. A terminal constraints-based optimization method is then employed to account for the diverse motion states of surrounding vehicles during the motion planning of the ego vehicle. Additionally, the multi-objective optimization problem of obstacle avoidance is reformulated as a control problem with well-defined objectives. Then, the performance of super-twisted sliding mode control (STSMC) and model predictive control (MPC) across different obstacle avoidance distances is subsequently compared. Finally, the proposed framework is validated on a hardware-in-the-loop (HIL) platform. The results demonstrate that the decisionmaking planner, based on avoidance distance, provides collisionfree navigation for industrial vehicles in complex scenarios. It also accurately categorizes obstacle avoidance challenges, significantly enhancing both vehicle stability during avoidance and the computational efficiency of the system.