An Advanced Rider-Cornering-Assistance System for PTW Vehicles Developed Using ML KNN Method

Area of Science:

• Motorcycle dynamics and control systems
• Artificial Intelligence in vehicle safety
• Machine Learning for predictive modeling

Background:

• Powered Two-Wheeler (PTW) dynamics are complex due to coupled longitudinal and lateral motion, especially at large roll angles.
• Existing cornering assistance systems often use restrictive models that can limit vehicle performance and user acceptance.
• The need for advanced rider assistance systems that enhance safety without compromising the riding experience is critical.

Purpose of the Study:

• To develop an Advanced Rider-cornering Assistance System (ARAS) for PTWs.
• To create AI and Neural Network-based algorithms that learn rider skills for curvilinear trajectories.
• To improve the estimation of cornering velocity using Machine Learning techniques.

Main Methods:

• Utilized Artificial Intelligence (AI) and Neural Network (NN) techniques to model rider behavior on curved paths.
• Developed new algorithms based on the K-Nearest Neighbor (KNN) Machine Learning (ML) technique for velocity prediction.
• Trained models on rider data from curvilinear trajectories to learn dynamic behaviors.

Main Results:

• Achieved high prediction accuracy of up to 99.06% for cornering velocity estimation.
• The developed ARAS system aims to provide assistance without imposing restrictive limitations on the rider.
• Demonstrated the potential of AI and ML in creating adaptive rider assistance systems.

Conclusions:

• The ARAS system offers a promising approach to enhance PTW safety by accurately predicting cornering velocity.
• AI and ML techniques can effectively replicate rider skills, leading to more intuitive and less restrictive assistance systems.
• Future work can focus on integrating this system for real-time rider support, improving overall PTW safety and performance.