Answers are grounded in this paper and cite their sources.
Experimental evaluation demonstrates reduced computation time and improved trajectory smoothness in autonomous vehicles, indicating enhanced real-time obstacle avoidance.
Local path planning is essential for autonomous vehicles operating in complex non-convex environments. Existing geometric planners are computationally efficient but often generate curvature discontinuities and steering commands that are difficult to execute, whereas optimization-based methods usually require higher computational cost and may suffer from numerical oscillations in constrained spaces. To address these issues, this paper proposes a dynamic potential field-guided and force-coupled adaptive pure pursuit framework, termed i-APP, for real-time local path planning. The proposed method consists of three stages. First, a velocity-adaptive gradient A* search generates a collision-free initial path with speed-dependent safety margins. Second, a dynamic artificial potential field refines risky path segments through deterministic lateral offsets to improve obstacle clearance. Finally, a force-coupled adaptive pure pursuit strategy maps obstacle repulsive forces to the look-ahead target point, while forward kinematic integration and steering-rate constraints are incorporated to improve trajectory executability. Simulation and real-vehicle experiments are conducted to compare i-APP with APP, A*+APP, and EMplanner methods. The results show that i-APP reduces planning time while improving trajectory smoothness and suppressing curvature fluctuation, indicating its suitability for real-time local planning in obstacle-constrained autonomous driving scenarios.
No takes yet. Share an insight, caveat, or question.
Jing et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: