Can Genetic Algorithms Be Used for Real-Time Obstacle Avoidance for LiDAR-Equipped Mobile Robots?

Area of Science:

• Robotics
• Artificial Intelligence
• Computational Intelligence

Background:

• Mobile robot deployment in public spaces is limited by real-time navigation challenges.
• Robots require smooth trajectory planning to avoid static and dynamic obstacles.

Purpose of the Study:

• To investigate the feasibility of employing genetic algorithms for real-time obstacle avoidance in mobile robots.
• To develop and evaluate a novel algorithm combining genetic algorithms with the velocity obstacle model.

Main Methods:

• A new family of algorithms, GAVO (Genetic Algorithms for Velocity Obstacles), was developed.
• Genetic algorithms were adapted for online, real-time optimization, moving beyond their traditional offline use.
• Experiments were conducted to test the performance of GAVO with specific chromosome representations and parameter settings.

Main Results:

• The GAVO algorithm demonstrated the capability for real-time obstacle avoidance.
• Carefully selected chromosome representations and parametrization are crucial for achieving real-time performance.
• The study validates the potential of genetic algorithms in dynamic navigation tasks.

Conclusions:

• Genetic algorithms can be effectively utilized for real-time obstacle avoidance in mobile robotics.
• The GAVO approach offers a promising solution for enhancing robot navigation in complex, dynamic environments.
• Further research into GA parametrization can unlock broader applications in autonomous systems.