Constructing characteristic signals of head acceleration events in stock car racing

Drivers in stock car racing are exposed to head acceleration events (HAEs) during racing. Currently, no head kinematic characteristic signals have been developed in a motorsport setting. Furthermore, the variability of loading in motorsport HAEs may not be adequately accounted for using common corridor-generation approaches. To address these limitations, ARCGen was employed which uses arc-length reparameterization and signal registration to develop characteristic signals in a wide range of settings. The objective of this study was to characterize HAEs in a motorsport setting using generated characteristic signals.

Head kinematics were collected from 20 NASCAR Cup Series drivers over 41 races during the 2024 season using mouthpiece sensors. Each sensor was configured to collect linear and rotational head kinematics (1600 Hz, 4 g threshold). All HAEs were visually verified as either race events (i.e., during the race, not associated with a crash) or crash events (i.e., during the race, associated with a crash resulting in a caution). HAEs containing multiple head impacts exceeding the trigger threshold were segmented into individual impacts. For each individual impact, peak kinematics were calculated including peak resultant linear acceleration (PLA) and angular acceleration (PRA). Principal direction of force (PDOF) was calculated from linear velocity. Impacts were separately binned into kinematic percentiles by PLA magnitude and PRA magnitude. Impacts were grouped into distinct bins based on PDOF, kinematic percentile, and event type (race, crash) for each binning method. To account for the variability of loading in motorsport HAEs, ARCGen (an open-source software package) was used to generate characteristic signals using arc-length reparameterization and signal registration for each bin in the x, y, and z directions and x-y resultant, and x-y-z resultant signatures. Signals were generated for linear acceleration, linear velocity, rotational acceleration, and rotational velocity. All signals were aligned using the optimal warping of resultant linear acceleration. Magnitude and duration were compared using descriptive statistics and qualitative characteristics were visually analyzed. Duration was calculated using 30% and 50% thresholding methods and compared.

3,156 HAEs (n = 2,750 race, n = 406 crash) were recorded. These HAEs were segmented into 4,478 individual impacts (n = 3,818 race, n = 660 crash). The rear PDOF characteristic signal for each percentile bin exposed drivers to the highest PLA magnitude during race and crash impacts, whereas the right side PDOF signal showed the lowest PLA magnitude for race impacts. Across all kinematic percentiles for race impacts, the right side PDOF signals exhibited the longest duration when compared to all other PDOFs for both duration calculation methods.

Motorsport-specific HAE characteristic signals were created. These may inform driver safety including helmet testing standards, finite element model validation, and safety system enhancements. This work facilitates cross-sport comparisons to contextualize motorsport racing HAEs relative to other environments.