Cyber Racing Coach: A Haptic Shared Control Framework for Teaching Advanced Driving Skills

TL;DR

This paper presents a haptic shared control framework for teaching advanced driving skills, integrating a real-time Model Predictive Control (MPC) autonomously planning trajectories with a haptic torque interface to guide learning. A fading scheme gradually reduces autonomy assistance during training, enabling long-term skill acquisition in high-performance driving tasks. A human-subject study comparing self-learning, full-assist, and fading-assist conditions shows that the fading approach yields superior skill retention, consistency, and safety on a racing track. The work demonstrates the potential of physics-based autonomy and tactile guidance to train drivers to push vehicle handling toward its dynamic limits, with implications for both high-performance contexts and everyday safety training. Future directions include adaptive fading, online human-modeling, and integration of additional feedback modalities to further enhance transfer and retention.

Abstract

This study introduces a haptic shared control framework designed to teach human drivers advanced driving skills. In this context, shared control refers to a driving mode where the human driver collaborates with an autonomous driving system to control the steering of a vehicle simultaneously. Advanced driving skills are those necessary to safely push the vehicle to its handling limits in high-performance driving such as racing and emergency obstacle avoidance. Previous research has demonstrated the performance and safety benefits of shared control schemes using both subjective and objective evaluations. However, these schemes have not been assessed for their impact on skill acquisition on complex and demanding tasks. Prior research on long-term skill acquisition either applies haptic shared control to simple tasks or employs other feedback methods like visual and auditory aids. To bridge this gap, this study creates a cyber racing coach framework based on the haptic shared control paradigm and evaluates its performance in helping human drivers acquire high-performance driving skills. The framework introduces (1) an autonomous driving system that is capable of cooperating with humans in a highly performant driving scenario; and (2) a haptic shared control mechanism along with a fading scheme to gradually reduce the steering assistance from autonomy based on the human driver's performance during training. Two benchmarks are considered: self-learning (no assistance) and full assistance during training. Results from a human subject study indicate that the proposed framework helps human drivers develop superior racing skills compared to the benchmarks, resulting in better performance and consistency.

Figures (5)

• Figure 1: The framework of shared control. Solid arrow represents real-time information transfer, while hollow arrow indicates information transfer only at the beginning of each run. (A1) 3 Degree of Freedom (DoF) bicycle model (see Sec. \ref{['sec:VD']}). (A2) Hard constraint of friction circle limit. (A3) Surface plot of the potential field for the cost of boundary violation (see Sec. \ref{['sec:SC']}). (B) Plant model for simulation. (C) Visualization in Unreal Engine 4 (see Sec. \ref{['sec:exp_setup']}). (D1) Logitech G29 steering wheel with torque sensor (see Sec. \ref{['sec:exp_setup']}). (D2) Human subject (see Sec. \ref{['sec:human_subject']}). (D3) Schematic block diagram of steering torque generator. (E) Evolution of assistance level along the trial number of one subject guided by fading scheme (see Sec. \ref{['sec:FS']}). (F) Design of the three gains $\alpha, \beta, \gamma$ as functions of the assistance level
• Figure 2: Experimental Setup: (A) The simulation with a human driver in the loop. (B) The torque sensor added to the steering column. (C) The track map of Thunderhill West Raceway and the autonomy path without human intervention. (D) The user interface to control the workflow of the testbed and display necessary information to the human driver between trials.
• Figure 3: Racing scores and success rates of pre-training tests. The black dots represent the real data from individual subjects. The height of each bar indicates the mean value, and the error bars depict the standard error. The brackets above the bar chart indicate the level of statistical significance.
• Figure 4: Assistance level and racing score throughout the training phase. The center line indicates the mean value of the metric, while the upper and lower lines denote one standard deviation above and below the mean, respectively.
• Figure 5: Post-training results. (A) Statistical analysis of metrics in the post-training test. (B) Example vehicle paths. Middle line is the mean path on the track and the two lines bounding the middle line denote one standard deviation from the mean path. The grey area magnified the driver performance in turn 7.