Neuro-fuzzy controller to navigate an unmanned vehicle

Area of Science:

• Robotics
• Artificial Intelligence
• Control Systems

Background:

• Autonomous vehicle navigation relies heavily on precise control strategies.
• Traditional methods like Artificial Neural Networks (ANNs) lack interpretability.
• Fuzzy Logic Controllers offer rule-based interpretability but can be difficult to tune.

Purpose of the Study:

• To develop and implement a novel neuro-fuzzy control method for Unmanned Vehicle (UV) navigation.
• To enhance the autonomous navigation ability and road-following precision of UVs.
• To leverage the complementary strengths of Neural Networks and Fuzzy Logic for improved control.

Main Methods:

• Implementation of an Artificial Neural Network Fuzzy Inference System (ANFIS) controller.
• Simulation of an autonomous vehicle navigating a complex environment with terrain and obstacles.
• Comparison of ANFIS performance against traditional Artificial Neural Network (ANN) methods.

Main Results:

• The neuro-fuzzy control system demonstrated significant improvements in control performance.
• Enhanced navigation ability and road-following precision were observed compared to previous methods.
• The ANFIS controller effectively integrated learning capabilities with interpretable fuzzy logic rules.

Conclusions:

• Neuro-fuzzy techniques, specifically ANFIS, offer a superior approach to autonomous vehicle control.
• Combining Neural Networks and Fuzzy Logic provides a robust and interpretable control solution.
• This method enhances UV performance in complex, dynamic environments.