Mobile Robot Localization Based on the PSO Algorithm with Local Minima Avoiding the Fitness Function

Area of Science:

• Robotics
• Artificial Intelligence
• Optimization Algorithms

Background:

• Accurate mobile robot localization is crucial for navigation.
• Existing methods like odometry and standard optimization algorithms can suffer from errors due to environmental disturbances and local minima.
• Semi-humanoid mobile robot Pepper's localization needs robust solutions for real-world applications.

Purpose of the Study:

• To propose a novel Particle Swarm Optimization with an Avoiding Local Minima (PSO-ALM) algorithm for robust mobile robot localization.
• To enhance localization accuracy by preventing the algorithm from getting trapped in local minima.
• To validate the proposed method through simulations and real-world experiments against various benchmark algorithms.

Main Methods:

• Development of a novel fitness function for PSO-ALM that includes integrity checks between consecutive solutions.
• Implementation of the PSO-ALM algorithm for the localization of the Pepper mobile robot using LIDAR data.
• Comparative analysis against standard Particle Swarm Optimization (PSO), Odometry, Genetic Algorithm (GA), GA-ALM, Golden Eagle Optimization (GEO), and GEO-ALM.

Main Results:

• The PSO-ALM algorithm demonstrated robust performance against disturbance perturbations in LIDAR measurements.
• PSO-ALM significantly improved localization accuracy compared to standard PSO and provided comparable or better results than GA and GEO algorithms, especially after initial travel.
• The Avoiding Local Minima (ALM) extension proved effective in preventing optimization algorithms from getting stuck in local minima, with PSO-ALM yielding the best overall results.

Conclusions:

• The proposed PSO-ALM methodology enables accurate and real-time global localization for mobile robots.
• The novel fitness function effectively prevents the algorithm from misidentifying the robot's location by penalizing nonsensical solutions.
• PSO-ALM offers a consistent and predictable time complexity, making it suitable for practical robotic applications.