Motorcycle steering torque estimation using a simplified front assembly model: experimental validation and manoeuvrability implications

Steering torque constitutes the primary motorcycle control input for the lateral dynamics; consequently, estimating it is important. Conventionally, this is done with complete motorcycle models, requiring significant identification effort. The simplified models in the literature only describe the steering torque under specific cases. This work defined a steering assembly model with few parameters to estimate the steering torque analytically for stationary and transient manoeuvres. The model equations followed from existing motorcycle models through simplifying hypotheses; transfer functions describing the roll response and the Lane Change Roll Index (LCRI) were obtained from these equations. Measured steering torque signals from different datasets, including diverse motorcycle classes, were used as the reference for validation. A good agreement resulted between the estimated and measured torques, in the time and speed-acceleration domains and in terms of LCRI. When using the roll as the motorcycle response, manoeuvrability was highest at lower frequencies. The scooter was the most manoeuvrable; the sports and touring motorcycles were the least manoeuvrable at low and high frequencies, respectively. Concerning design parameters, the front-wheel spin inertia and front twist stiffness influenced manoeuvrability the most. The model allows recreating the steering torque signal for new and pre-existing datasets using commonly measured signals; the signal can describe the riding style and the effort required. Few parameters are required, facilitating its use and reducing the computational burden, allowing its use for steering assistance systems.