Development and validation of a friction estimation model for collision avoidance manoeuvres in autonomous trucks

Abstract

Autonomous trucks are rapidly gaining interest in the commercial vehicle sector due to their potential to improve road safety, reduce operational costs, and optimize long-haul transport. However, one of the critical challenges in ensuring the safety of these vehicles lies in their ability to perform effective collision avoidance manoeuvres, especially under varying road surface conditions and near the tire-road friction limits. Accurate knowledge of the available friction is essential for making safe and optimal decisions regarding braking and steering during emergency situations. This thesis presents the development and validation of a real-time road-tire friction estimation model designed specifically for autonomous truck applications. The proposed estimator leverages a longitudinal vehicle dynamics-based approach, using the slip-slope method to estimate the tire-road friction coefficient. A recursive least squares (RLS) algorithm is employed to update the friction estimate in real time based on inputs such as wheel slip, normal load, and longitudinal acceleration extracted from truck test log data.The estimated peak friction coefficient is used by a collision avoidance controller that dynamically selects between braking-only and sequential brake-and-steer strategies. The controller incorporates constraints on maximum lateral acceleration and steering angle to ensure stability and safety. Simulation studies were performed using MATLAB to validate the integrated system, and physical tests were conducted with a Scania 4x2 tractor truck to compare the predicted trajectories with real-world behaviour.