Data-Driven Methodology for the Investigation of Riding Dynamics: A Motorcycle Case Study

Powered-Two-Wheelers (PTW) riders’ fatalities are prevalent on bends outside built-up areas due to the complexity and instability of their vehicles: countermeasures require a better understanding rider-PTW interaction. Analysing riding data is effective but becomes challenging when using extensive datasets; segmenting would help identify events interest, isolate specific manoeuvres describe session. Manual segmentation be time-consuming subjective; automation beneficial. This work proposed an automatic, unsupervised tool for clustering signals acquired during session studying motorcycle lateral dynamics in-depth. The method only requires measuring roll angle. An expert rider completed closed route instrumented motorcycle; algorithm divided time series into segments categorised clusters relative conditions. segmented trial revealed effectiveness usefulness approach. Then, corner entry manoeuvre was investigated in-depth observe each segment’s properties. associated primitive cluster described through segments’ succession. were unambiguous easy interpret thanks dynamics-based nature minimal overlap. identified differences between three in trial. simplified analysis allowed early detection start. can aid research dynamics, PTW-rider interaction, preferences bends. could employed training pre-crash fall reconstruction.