Human-machine shared control vehicles (HSCVs) keep the driver and the automation simultaneously in the control loop, with control authority dynamically allocated between them. Zhou, Huang and Hang (Applied Sciences 2024) proposed a game-theoretic interactive control method in which a BELCM-based driver surrogate handles steering and speed, and the HSCV interacts with multiple surrounding vehicles through a non-cooperative game solved by distributed model predictive control. Here BELCM (brain emotional learning circuit model) is a computational model mimicking the brain's emotional learning process, used as a human-like driver surrogate. The validation, however, rested on scripted assumptions about the driver's lane-change and deceleration intentions, and the authors explicitly leave open the extension to a more realistic integrated driver model that captures steering and speed behavior together with driver intention, and especially the impact of driver intention on shared control. The surrounding literature shows that these halves have progressed separately — integration without intention, or intention without integration or interactive games — so the coupled core remains open.

Background

Shared control is a driving paradigm in which the human driver and an automation system remain in the control loop at the same time and jointly complete the driving task; it is widely viewed as a transitional stage toward full automation because it combines the machine's superiority in sensing and computation with the human's superiority in judgment. A human-machine shared control vehicle (HSCV) blends the driver's commands with the automation's commands through an authority-allocation mechanism that decides, at each instant, how much control each agent exerts. Two long-standing gaps in this field are relevant here. First, most shared-control studies address only lateral (steering) control under constant speed, so integrated frameworks that treat steering and speed control together are comparatively rare. Second, the driver's intention — what maneuver the driver is trying to perform, such as changing lanes or decelerating — is a latent mental state that must be inferred from observable signals; many controllers simply assume it, which limits realism whenever the assumption is wrong.

Zhou, Huang and Hang (2024) proposed a safety-oriented interactive control method for HSCVs in multi-vehicle traffic. Its ingredients are: a driver model, based on the brain emotional learning circuit model (BELCM — a computational model mimicking the brain's emotional learning process, used here as a human-like controller surrogate), that integrates steering control and speed control; a shared-control strategy that dynamically allocates driving control authority between the driver and the automation according to a driving safety field (a potential-field measure of environmental risk) and the vehicle speed; a characterization of surrounding vehicles' uncertain behavior through aggressiveness levels built from driving-safety and travel-efficiency indicators; and a formulation of the interaction between the HSCV and the surrounding vehicles as a non-cooperative game, solved approximately as a distributed model predictive control (DMPC) problem — model predictive control being an optimization-based control method that repeatedly solves a finite-horizon optimization, and the distributed variant letting each vehicle optimize its own cost while anticipating the others, with the solution interpreted as a Nash equilibrium (a state where no player gains by unilaterally changing strategy). The driver's control input enters this solution process. Two simulation scenarios (lane keeping and lane changing with neighbor vehicles of different aggressiveness) showed that the automation can promptly override dangerous driver operations and safely interact with multiple vehicles.

The limitation the authors themselves flag is that the simulated driver is a BELCM surrogate and that the test scenarios rest on 'reasonable assumptions about the driver's intention to change lanes and deceleration intention': intention is scripted, not modeled or estimated, and therefore cannot influence, or be influenced by, the authority allocation or the game interaction. Their conclusion states that 'a more realistic integrated driver model that considers steering control and speed control requires us to conduct further research in the future, especially on the impact of driver intention on shared control, to improve shared control technology in HSCVs.' Neighboring literature makes the shape of the missing piece concrete: intention recognition has been studied with sequence models such as double-layer LSTM (long short-term memory) networks and CNN-GRU (convolutional neural network with gated recurrent unit) predictors, typically for steering alone or for takeover events; integrated lateral-longitudinal authority allocation has been achieved with game-theoretic and robust optimization methods but without any representation of driver intention; and the interaction with surrounding vehicles of different driving styles is precisely the setting in which a misread intention is most safety-critical.

Problem Statement

Extend game-theoretic shared control of human-machine cooperative vehicles (HSCVs — vehicles in which the driver and the automation remain in the control loop together, with control authority dynamically allocated between them) to a realistic integrated driver model that explicitly represents driver intention, and determine how driver intention affects shared control. Relative to the framework of Zhou, Huang and Hang (2024), in which the driver is simulated by a surrogate based on the brain emotional learning circuit model (BELCM — a computational model mimicking the brain's emotional learning process, used as a human-like controller) and lane-change and deceleration intentions are scripted test-scenario assumptions, an answer must deliver three jointly linked results: (i) an integrated driver model coupling steering (lateral) and speed (longitudinal) control behavior with an explicit representation of driver intention — the maneuver the driver is attempting, a latent driver state in the sense of the Background — in place of an intention preset for the whole scenario; the intention may be carried as an internal state of the driver model or estimated online from driver observables such as steering torque, pedal activity and vehicle state, but on either route it must enter the shared-control loop as a run-time variable rather than a per-scenario script; (ii) the embedding of this model in the shared-control loop so that this intention signal influences the dynamic allocation of control authority between driver and automation and the HSCV's non-cooperative interaction with surrounding vehicles, together with a characterization of that influence — including, for example, how authority allocation and the interactive strategy change when the intention is unsafe, time-varying, or in conflict with the automation's risk assessment (it suffices to characterize at least one of these three conditions, with the others optional strengthening); and (iii) an evaluation of the extended framework under richer and more realistic human driving intentions and interaction conditions than the two assumed scripts (it suffices to demonstrate the intention effect in simulation with at least one run-time intention signal and surrounding vehicles of at least two different driving styles; driver-in-the-loop experiments and further driving styles are optional strengthening), such that the claimed effect of driver intention on shared control is demonstrated by the behavior of the extended framework rather than assumed in the scenario design. These components serve one overarching objective — intention-aware integrated shared control for human-machine cooperative driving — and the problem is complete only when the model, its integration into the interactive shared-control framework, and the evidence of the intention effect are all present.

References

  1. Yangyang Zhou, Chao Huang, Peng Hang. Game Theory-Based Interactive Control for Human-Machine Cooperative Driving. Applied Sciences, 14(6):2441, 2024. DOI: 10.3390/app14062441. https://doi.org/10.3390/app14062441
  2. Yangyang Zhou, Jingrui Guo, Dong Hu, Pai Zheng, Chao Huang. Safe Driving for Human-Machine Shared Control: A Prediction-Informed Risk-Aware SMPC. 2026 IEEE International Conference on Human-Machine Systems (ICHMS), pp. 216-221, 2026. DOI: 10.1109/ICHMS69701.2026.11602158. https://doi.org/10.1109/ICHMS69701.2026.11602158
  3. Yingkui Shi, Sicheng Ge, Anh-Tu Nguyen, Ji Li, Chao Huang, Chuan Hu. Trust-Based Shared Steering Control With Game Theory Using Event-Triggered ADP for Human-Vehicle Cooperative Driving. IEEE Transactions on Vehicular Technology, 74(11):16879-16893, 2025. DOI: 10.1109/TVT.2025.3581082. https://doi.org/10.1109/TVT.2025.3581082
  4. Abhimanyu Pratap Singh, Husain Kanchwala, Kai Li Lim, Mehmet Yildirimoglu. Driving Together: Intent-Aware Dynamic Authority Allocation for Shared Control. SSRN preprint, 2026. DOI: 10.2139/ssrn.6459553. https://doi.org/10.2139/ssrn.6459553
  5. Yutao Chen, Jun Cheng, Hongliang Zhang, Jie Huang. Robust and dynamic control authority allocation in human-machine shared vehicle control: An integrated lateral-longitudinal optimization approach. Control Engineering Practice, 175:107073, 2026. DOI: 10.1016/j.conengprac.2026.107073. https://doi.org/10.1016/j.conengprac.2026.107073
  6. Wenlong Ding, Xiang Chen, Wanzhong Zhao, Chunyan Wang. Lateral and Longitudinal Shared Control Strategy of Human-Machine Based on Game-Theoretic. Journal of Physics: Conference Series, 2832:012020, 2024. DOI: 10.1088/1742-6596/2832/1/012020. https://doi.org/10.1088/1742-6596/2832/1/012020
  7. Zhaoqing Wang, Jin Huang, Yuhang Xu, Xiaozhou Wu, Xueyun Li, Mengmeng Yang, Zhihua Zhong. A Human-Machine Shared Control Strategy for Automation Disengagement Cases with Intention-Based Driving Authority Allocation. 2024 IEEE 27th International Conference on Intelligent Transportation Systems (ITSC), pp. 3807-3814, 2024. DOI: 10.1109/ITSC58415.2024.10919712. https://doi.org/10.1109/ITSC58415.2024.10919712
  8. Weida Wang, Yunpu Zhang, Chao Yang, Yuhang Zhang, Yipeng Gao, Taiheng Ma, Tianqi Qie. A human-machine shared dual fuzzy authority allocation control strategy for automatic driving vehicle considering driver intention judgement. Expert Systems with Applications, 274:126971, 2025. DOI: 10.1016/j.eswa.2025.126971. https://doi.org/10.1016/j.eswa.2025.126971
  9. Chen Zhou, Fan Zhang, Edric John Cruz Nacpil, Zheng Wang, Fei-Xiang Xu. Driver Steering Intention Prediction for Human-Machine Shared Systems of Intelligent Vehicles Based on CNN-GRU Network. Sensors, 25(10):3224, 2025. DOI: 10.3390/s25103224. https://doi.org/10.3390/s25103224
  10. Junhui Zhang, Xiaoman Guo, Zongjie Fu, Yuxi Liu, Yuxuan Ding. Noncooperative-Game-Theory-Based Driver-Automation Shared Steering Control Considering Driver Steering Behavior Characteristics. IEEE Internet of Things Journal, 11(17):28465-28479, 2024. DOI: 10.1109/jiot.2024.3401120. https://doi.org/10.1109/jiot.2024.3401120