Kinematic/Dynamic SLAM for Autonomous Vehicles Using the Linear Parameter Varying Approach.

Area of Science:

• Robotics
• Control Systems
• Artificial Intelligence

Background:

• Traditional mobile robotics SLAM often uses kinematic models.
• Autonomous vehicles require dynamic models due to mass, velocity, and effects like slip angle.
• Kinematic models are insufficient for accurate autonomous vehicle localization.

Purpose of the Study:

• To develop an improved SLAM-based localization system for autonomous vehicles.
• To address the limitations of kinematic models by incorporating vehicle dynamics.
• To leverage advanced control techniques for nonlinear model handling.

Main Methods:

• Augmenting SLAM with both kinematic and dynamic vehicle behaviors.
• Employing Linear Parameter Varying (LPV) techniques to manage model nonlinearities.
• Designing state estimation schemes using systematic LPV methodologies.

Main Results:

• The proposed LPV-based SLAM system effectively handles nonlinearities in autonomous vehicle models.
• The system demonstrates superior localization performance compared to classical Extended Kalman Filter (EKF) methods.
• LPV techniques provide a robust framework for advanced state estimation in dynamic systems.

Conclusions:

• Integrating dynamic vehicle behavior into SLAM is crucial for autonomous vehicles.
• LPV techniques offer a powerful approach to managing complex nonlinearities in robotic systems.
• The developed system provides more accurate and reliable localization for autonomous vehicles.