Improved RRT*-Connect Manipulator Path Planning in a Multi-Obstacle Narrow Environment.

Area of Science:

• Robotics
• Artificial Intelligence
• Motion Planning

Background:

• Robotic arm path planning in narrow, obstacle-rich environments presents significant computational challenges.
• Existing algorithms often struggle with efficiency and optimality in complex scenarios.

Purpose of the Study:

• To propose an improved Rapidly-exploring Random Tree star-Connect (IRRT*-Connect) algorithm for enhanced robotic arm path planning.
• To address limitations in sampling efficiency, search acceleration, and path optimality in cluttered environments.

Main Methods:

• Implemented a heuristic sampling strategy integrating ellipsoidal subset and goal-biased sampling to compress the search space.
• Introduced an adaptive step-size and node rejection strategy for accelerated, collision-free path generation.
• Applied cubic non-uniform B-spline interpolation for path smoothing and pruning optimization to remove redundant nodes.

Main Results:

• IRRT*-Connect demonstrated reduced planning times compared to existing algorithms.
• The proposed method achieved lower path costs for robotic manipulator operations.
• Successful simulations in Python/ROS and physical experiments on a UR5 robotic arm validated the approach.

Conclusions:

• The IRRT*-Connect algorithm offers a significant improvement for robotic arm path planning in challenging environments.
• The enhanced sampling and search strategies contribute to faster and more optimal path generation.
• This work provides a robust solution for real-world robotic applications requiring efficient motion planning.