Accelerating RRT* convergence with novel nonuniform and uniform sampling approach.

Area of Science:

• Robotics
• Artificial Intelligence
• Computer Science

Background:

• Path planning is essential for autonomous mobile robots.
• Sampling-based algorithms like RRT* are common for collision-free path generation.
• RRT* suffers from slow convergence due to uniform sampling.

Purpose of the Study:

• To introduce a novel hybrid sampling method, RRT*-NUS (nonuniform-uniform sampler).
• To enhance the exploration efficiency of sampling-based path planners.
• To improve convergence speed and reduce planning time for autonomous robots.

Main Methods:

• Proposed a hybrid sampling strategy combining nonuniform and uniform sampling (RRT*-NUS).
• Evaluated RRT*-NUS against six baseline algorithms (RRT*, Informed RRT*, RRT*-N, GS-RRT*, DR-RRT*, hybrid-RRT*).
• Conducted simulations in three 384*384 2D environments.

Main Results:

• RRT*-NUS demonstrated superior performance over baseline RRT* algorithms.
• Achieved a 67.5% improvement in planning time compared to RRT*.
• Reached a convergence rate of 0.41 units/s, significantly faster than RRT* and Hybrid RRT*.

Conclusions:

• The RRT*-NUS method offers a significant advancement in autonomous robot path planning.
• Hybrid sampling effectively addresses the slow convergence issue of traditional RRT*.
• RRT*-NUS provides a more efficient and faster solution for generating collision-free paths.