Graph-based Path Planning with Dynamic Obstacle Avoidance for Autonomous Parking

TL;DR

The paper tackles real-time, collision-free path planning for autonomous parking in environments with static and dynamic obstacles, noting the problem’s $NP$-hard nature under non-holonomic dynamics. It advances a time-indexed Hybrid A* algorithm that directly integrates predictions of dynamic obstacles into the search, complemented by an online planning framework that selects adaptive intermediate goals along a precomputed static-path heuristic. The main contributions are (1) a time-indexed Hybrid A* variant that leverages dynamic obstacle trajectories, (2) an online planner using adaptive look-ahead to compute locally feasible paths while following a global path, and (3) extensive simulations showing substantial improvements in computational efficiency and safety over a spline-based planning method across perpendicular, angle, and parallel parking scenarios and in densely populated lots. The work demonstrates the practicality of real-time, dynamic-obstacle-aware parking path planning for large-scale parking environments.

Abstract

Safe and efficient path planning in parking scenarios presents a significant challenge due to the presence of cluttered environments filled with static and dynamic obstacles. To address this, we propose a novel and computationally efficient planning strategy that seamlessly integrates the predictions of dynamic obstacles into the planning process, ensuring the generation of collision-free paths. Our approach builds upon the conventional Hybrid A star algorithm by introducing a time-indexed variant that explicitly accounts for the predictions of dynamic obstacles during node exploration in the graph, thus enabling dynamic obstacle avoidance. We integrate the time-indexed Hybrid A star algorithm within an online planning framework to compute local paths at each planning step, guided by an adaptively chosen intermediate goal. The proposed method is validated in diverse parking scenarios, including perpendicular, angled, and parallel parking. Through simulations, we showcase our approach's potential in greatly improving the efficiency and safety when compared to the state of the art spline-based planning method for parking situations.

Figures (7)

• Figure 1: A parallel parking scenario with static cars (black), pedestrians, and a moving car (red). The brown car is the ego vehicle. The more transparent images of the red car and the pedestrian denote their respective predictions.
• Figure 2: Geometry of the vehicle and obstacle avoidance. The vehicle is the brown rectangle and obstacle is the red circle.
• Figure 3: Target scenario for Problem \ref{['prob_1']}. The black rectangles are static vehicles, and the dynamic obstacle moves from right to left.
• Figure 4: Target scenario for Problem \ref{['prob_2']} with four dynamic obstacles where the local path of an intermediate time step is shown. The light brown rectangles denote the trajectory of the ego vehicle, and the black arrows denote the motion of dynamic obstacles.
• Figure 5: Comparison of paths generated by our time-indexed Hybrid A* method (t-HA* + A*) and the iterative spline-based method (ItCA).

Theorems & Definitions (2)

• Remark 1
• Definition 1