Multi-strategy cooperative enhancement dung beetle optimizer and its application in obstacle avoidance navigation

Area of Science:

• Robotics and Artificial Intelligence
• Optimization Algorithms
• Autonomous Systems

Background:

• Path planning is essential for autonomous navigation in unmanned driving systems.
• The standard dung beetle optimization algorithm (DBO) suffers from premature convergence and limited global planning.
• Addressing these limitations is critical for advancing autonomous driving capabilities.

Purpose of the Study:

• Introduce a novel multi-strategy cooperative enhanced dung beetle optimization algorithm (RCDBO).
• Enhance the DBO algorithm to overcome local optima and improve global path planning.
• Evaluate the effectiveness of RCDBO for autonomous navigation path planning.

Main Methods:

• Implemented Bernoulli-based chaotic mapping for improved initial population diversity.
• Incorporated a random walk strategy to prevent early-stage local stagnation.
• Utilized a vertical and horizontal crossover strategy to boost late-stage global optimization.
• Validated RCDBO against benchmark functions, CEC2021 test suite, and statistical tests.

Main Results:

• RCDBO demonstrated superior performance compared to other swarm intelligence algorithms on benchmark tests.
• The algorithm effectively balances global exploration and local exploitation.
• Path planning simulations showed RCDBO yields shorter path lengths than basic DBO.
• RCDBO produced smoother paths with fewer turns in simpler environments.

Conclusions:

• RCDBO effectively addresses the limitations of the basic DBO algorithm for path planning.
• The proposed enhancements significantly improve global optimization and prevent local optima.
• RCDBO offers a promising solution for path planning in unmanned driving systems, enhancing efficiency and path quality.