Path Planning for Dragon-Fruit-Harvesting Robotic Arm Based on XN-RRT* Algorithm

Area of Science:

• Robotics
• Artificial Intelligence
• Agricultural Engineering

Background:

• Path planning for robotic harvesting in complex environments faces challenges with efficiency and success rates.
• Existing algorithms like RRT* may struggle with suboptimal solutions and convergence in dynamic agricultural settings.

Purpose of the Study:

• To develop and evaluate an enhanced Rapidly-exploring Random Tree star (RRT*) algorithm, termed XN-RRT*, for improved robotic arm path planning in pitaya harvesting.
• To address issues of low path planning efficiency and suboptimal picking success rates in complex harvesting scenarios.

Main Methods:

• Proposed XN-RRT* algorithm utilizing normal distribution for sampling points with dynamic adjustments based on target distance and tree density.
• Incorporated an improved artificial potential field method during tree expansion to guide sampling and overcome local optima.
• Employed a greedy algorithm for node reduction and cubic B-spline curves for path smoothing.

Main Results:

• Simulations in 2D and 3D environments showed XN-RRT* achieved higher convergence efficiency and superior path quality with fewer iterations.
• In a simulated pitaya orchard, XN-RRT* achieved a 98% picking path planning success rate, significantly outperforming RRT*.
• The algorithm demonstrated a 27.12% reduction in path length and a 14% increase in planning success rate compared to RRT*.

Conclusions:

• The XN-RRT* algorithm offers excellent overall performance for robotic arm path planning in complex agricultural harvesting environments.
• The enhanced sampling and guidance strategies effectively reduce redundant sampling and improve path optimality.
• Results provide a valuable reference for advancing robotic harvesting technologies and path planning solutions.