Autonomous Emergency Braking With Driver-In-The-Loop: Torque Vectoring for Active Learning

TL;DR

The paper tackles the challenge of reliable autonomous emergency braking (AEB) under uncertain tyre-road friction by online identification of the current and maximum friction coefficients, leveraging Torque Vectoring for Active Learning (TVAL) within a driver-in-the-loop framework. It combines Dual Control for Exploration and Exploitation (DCEE) with a Regularized Particle Filter to estimate the Magic Formula tyre parameters $\Theta=[B,C,D,E]^T$ and augmented states $\mathbf{\chi}=[v,\omega_f,\omega_r,\Theta]^T$, while using a rain/light sensor to trigger resampling and a hysteresis-based activation to balance learning with driver control. The approach demonstrates substantial reductions in estimation uncertainty and energy consumption on a high-fidelity 7-DOF vehicle model across dry, wet, and snow road surfaces, achieving safe transitions between active learning and driver input and enabling more reliable preemptive braking decisions. These results advance online safety-critical control in AEB and offer transferable insights for parameter identification in robotics and drive-by-wire systems.

Abstract

Autonomous Emergency Braking (AEB) potentially brings significant improvements in automotive safety due to its ability to autonomously prevent collisions in situations where the driver may not be able to do so. Driven by the poor performance of the state of the art in recent testing, this work provides an online solution to identify critical parameters such as the current and maximum friction coefficients. The method introduced here, namely Torque Vectoring for Active Learning (TVAL), can perform state and parameter estimation whilst following the driver's input. Importantly with less power requirements than normal driving. Our method is designed with a crucial focus on ensuring minimal disruption to the driver, allowing them to maintain full control of the vehicle. Additionally, we exploit a rain/light sensor to drive the observer resampling to maintain estimation certainty across prolonged operation. Then a scheme to modulate TVAL is introduced that considers powertrain efficiency, safety, and availability in an online fashion. Using a high-fidelity vehicle model and drive cycle we demonstrate the functionality of TVAL controller across changing road surfaces where we successfully identify the road surface whenever possible.

Figures (18)

• Figure 1: TVAL system architecture for AEB with the novel contribution shown in red. A Global Navigation Satellite System (GNSS) e.g. GPS gives vehicle speed measurements and Dual Control for Exploration-Exploitation (DCEE) is used as part of our control strategy.
• Figure 2: 7 DOF vehicle dynamics model.
• Figure 3: Wheel slip - friction coefficient relationship for positive and negative slips.
• Figure 4: Identification of both current friction coefficient (top) and maximum available friction (bottom) over a varying road surface shown by the change in true maximum friction coefficient (red). Note that the blue curve is the mean value of the estimation, while the green and purple curves show the upper bound and the lower bound of the estimation.
• Figure 5: Comparison of different TVAL activation strategies. Results using sensor $S_{y,k}^2$ and estimation $S_{k}^2$ uncertainty are shown in green, error between sensor measurement and estimation is compared against a threshold of $1$ in pink and the result using estimation and prediction uncertainty is shown in blue.

Theorems & Definitions (4)

• Remark 1
• Remark 2
• Remark 3
• Remark 4