Local Path Planning with Dynamic Obstacle Avoidance in Unstructured Environments
🤖 AI Summary
This work addresses the local path planning problem for unmanned ground vehicles (UGVs) operating in unstructured environments with dense, dynamically moving obstacles. We propose a real-time collision avoidance algorithm that integrates tangent-path planning with polynomial trajectory extrapolation. Given a global path defined by waypoints, the method models dynamic obstacles’ motions as stochastic polynomial trajectories and combines tangent-based geometric avoidance with precise collision prediction to enable smooth, local path replanning. Our key contribution lies in the deep coupling of geometric tangent-path generation with dynamic obstacle trajectory extrapolation, significantly improving both computational efficiency and path continuity. Simulation results demonstrate that the algorithm reliably generates collision-free trajectories under complex multi-obstacle scenarios with stochastic motion, successfully guiding the UGV from start to goal with high success rate, minimal path oscillation, and computation time satisfying real-time constraints.
Application Category
📝 Abstract
Obstacle avoidance and path planning are essential for guiding unmanned ground vehicles (UGVs) through environments that are densely populated with dynamic obstacles. This paper develops a novel approach that combines tangent-based path planning and extrapolation methods to create a new decision-making algorithm for local path planning. In the assumed scenario, a UGV has a prior knowledge of its initial and target points within the dynamic environment. A global path has already been computed, and the robot is provided with waypoints along this path. As the UGV travels between these waypoints, the algorithm aims to avoid collisions with dynamic obstacles. These obstacles follow polynomial trajectories, with their initial positions randomized in the local map and velocities randomized between 0 and the allowable physical velocity limit of the robot, along with some random accelerations. The developed algorithm is tested in several scenarios where many dynamic obstacles move randomly in the environment. Simulation results show the effectiveness of the proposed local path planning strategy by gradually generating a collision free path which allows the robot to navigate safely between initial and the target locations.
Problem
Research questions and friction points this paper is trying to address.
Develops local path planning for unmanned ground vehicles in dynamic environments
Avoids collisions with obstacles moving along randomized polynomial trajectories
Generates collision-free paths between waypoints in unstructured environments
Innovation
Methods, ideas, or system contributions that make the work stand out.
Combines tangent-based planning with extrapolation methods
Generates collision-free paths in dynamic unstructured environments
Uses polynomial trajectories for dynamic obstacle modeling
🔎 Similar Papers
No similar papers found.
