Toward motorcycles’ autonomous emergency steering: methodological framework for testing the feasibility of interventions enabling crash avoidance

Objective: This research aimed to establish a methodological framework for field-testing motorcycle autonomous emergency steering (M-AES) safety function using a specialized prototype demonstrator to identify critical intervention parameters and propose objective metrics for quantifying vehicle stability and human-machine interaction in emergency maneuvers. Methods: The study was conducted using a standard street motorcycle equipped with a specifically developed prototype steering actuator capable of delivering controlled steering torques up to 30 Nm. M-AES intervention parameters were those indicated in prior work to be effective in preventing typical crash conditions. A data acquisition system recorded kinematic parameters at 100 Hz, including roll, roll rate, steering angle, and steering torque. The experimental protocol consisted of trials on a closed-to-traffic track testing the intervention of M-AES in different conditions: “Active M-AES” trials, where the rider allowed the system to operate autonomously and “passive M-AES” trials, where the rider’s task was to override M-AES action. To quantify the conflict between the rider and the system and the stability of the vehicle in correspondence with M-AES intervention, specific metrics were also developed. Results: First, this study established an experimental approach for the field testing of M-AES, building upon previous research on autonomous emergency systems for motorcycles and existing literature regarding effective crash-avoidance parameters. Field trials were conducted to validate the proposed methodology and the newly introduced metrics. Utilizing the M-AES prototype system, automated lane-change maneuvers were performed, achieving target roll angles of 25–35 deg and roll rates of 50 deg/s. Under “active M-AES” conditions, the vehicle remained stable across all tested speeds (40–60 km/h), with dynamic parameters confirming robust vehicle stability. Results from “passive M-AES” trials demonstrated that the motorcycle remained controllable during the autonomous intervention; riders successfully overrode the system’s torque by applying manual steering inputs, confirming the feasibility of rider-initiated overrides. Conclusions: This study established M-AES critical intervention parameters influencing motorcycle stability and objective metrics for quantifying human-machine interaction, providing a technical foundation for future large-scale human subject studies. This is the first testing protocol for evaluating M-AES and riders’ reciprocal actions using a prototype demonstrator on a standard motorcycle.