Effect of steering input form on motorcycle riding simulation model performance

In this study, we derived a motorcycle riding simulation model reproducing the steering input forms of Bruschetta et al. (2020) and Picotti et al. (2024), where the plant model input is the steering angle and the path-following controller (rider model) input is the steering rate, based on the riding simulation model of Hatakeyama et al. (2024b). For this purpose, a plant model was derived by modifying that of Hatakeyama et al. (2024a) and changing the steering torque input to steering angle input. Additionally, by removing the equation corresponding to the steering dynamics of the internal model of the path-following controller derived by Hatakeyama et al. (2024b), the steering torque input was changed to a steering rate input. Using the modified riding simulation model, along with the riding simulation model of Hatakeyama et al. (2024b), which adopted a steering input form where both the plant model and the path-following controller take the steering torque as the input, riding simulations were conducted under identical conditions. By keeping the model structure, controller weights, and solver of the nonlinear model predictive control identical, except for differences related to the steering input form, it was possible to focus the analysis on the impact of the steering input form on riding simulation performance. The riding simulation demonstrated that, in the simulation model with a steering input form where the plant model and path-following controller take the steering angle and rate as the inputs, the lack of steering dynamics resulted in excessive oscillations in the control input or even divergence of the response. This shows that in motorcycle model-based development, the response does not reflect reality unless the input to the plant model and path-following controller is the steering torque, as reported by Hatakeyama et al. (2024b).