Powered Two-Wheeler Riding Profile Clustering for an In-Depth Study of Bend-Taking Practices.

Area of Science:

• Road safety
• Motorcycle dynamics
• Human-machine interaction

Background:

• Understanding rider/vehicle interaction in motorcycle cornering is challenging due to limited instrumentation and diverse rider practices.
• Existing knowledge on motorcycle bend-taking strategies lacks comprehensive, data-driven insights into actual riding behaviors.

Purpose of the Study:

• To develop a data-driven methodology for identifying typical motorcycle rider behaviors during bend-taking using clustering.
• To enhance the understanding of powered two-wheeler (PTW) riding practices in curves through real-world data collection.

Main Methods:

• Collected real-world data on PTW dynamics (velocity, acceleration, jerk), road position (curvature), and handlebar input (steering angle) within the VIROLO++ project.
• Applied clustering methods to identify distinct riding behavior patterns in bends.
• Utilized Anderson-Darling tests for statistical analysis and compared clustering results with subjective rider data.

Main Results:

• Identified and contextualized typical riding tendencies in bends through clustering analysis.
• Demonstrated the effectiveness of a data-driven approach in characterizing motorcycle cornering behaviors.
• Highlighted differences in control strategies (handlebar steering vs. rider lean) via pressure sensor data analysis.

Conclusions:

• The developed methodology successfully categorizes motorcycle rider behaviors in bends.
• Data-driven analysis provides valuable insights into PTW cornering practices, complementing subjective rider accounts.
• Further analysis using pressure sensors can differentiate between steering and leaning control techniques.