考虑驾驶意图与动态环境的智能汽车人机驾驶切换控制方法研究

Intelligent vehicle autopilot technology is coming quickly to us, but the prevailing implementation into reality will take a long time. The short-term arrival will be a more intelligent driver-machine co-pilot system that performs dynamic driving tasks when the driver responds to a specific driving operation of the intervention request. Challenges existed when solving the problems of developing safety driving strategy and stable driving mode transition in complex scenes. The proposed research will start from the cognitive mechanism of risky driving behavior, and analyze and model the driving behavior characteristics and operation sequences under typical driving intention. By comprehensively evaluating the influence of the driving intention, vehicle motion and road traffic environment on the driving safety, a situational awareness model for fusing "human-car-road" multi-source information and describing the vehicle collision risk is constructed,based on which, specific method will be forwarded to resolve the vehicle motion control sovereignty distribution strategy and the intervention time of driving sovereign transition. The innovation of this proposal is to integrate the factors of driver intention and the dynamic traffic environment to establish an accurate model for examining risk driving situations in real-time, and then put forward the triggering condition and control method of the driver-machine driving switching strategy, and finally guarantee the safety and stability of driver-machine driving system. The research aims to improve the theoretical system of vehicle automatic driving technology in theory, and provides novel ideas for realizing the practical application of intelligent vehicle driver machine co-pilot system.

智能汽车自动驾驶技术正快速向我们走来，但实际道路中大规模实现还需较长时间。快速到来的将是更加智能的人机协同驾驶系统在驾驶员响应特定驾驶操作的干预请求时，执行动态驾驶任务。当前仍需重点解决复杂场景下的汽车安全驾驶策略，及驾驶控制主权的分配与切换方法。本研究将首先从汽车危险驾驶行为认知机理入手，对典型驾驶意图下的驾驶行为特征与操作序贯链进行分析与建模；综合评估驾驶意图、汽车运动状态和道路环境对行车安全性的影响；融合“人-车-路”多源信息构建汽车碰撞危险态势感知模型；在此基础上，提出汽车运动控制主权分配策略及主权切换的介入时刻界定。本项目创新点在于将驾驶意图与车外动态环境有机结合，建立准确的行车危险态势辨识模型，进而提出人机驾驶模式切换策略的触发条件和控制方法，确保人机驾驶切换过程的安全性和稳定性。研究成果在理论上可以完善汽车自动驾驶技术的理论体系，为实现智能汽车人机共驾的应用与实现提供新思路。

本项目围绕智能网联汽车行车危险态势感知，智能汽车自主驾驶和人工驾驶切换控制方法开展了深入研究。首先对典型驾驶意图下的驾驶员操作行为特征与操控序贯链进行了分析与建模；综合评估驾驶意图/驾驶行为、汽车运动状态以及动态环境对行车安全性的影响；通过对“人-车-路”多源异构信息进行深度融合和挖掘，构建了复杂道路交通环境下的汽车碰撞危险态势辨识模型；在此基础上，结合车辆运动状态和驾驶员个体差异性提出了人机驾驶模式切换策略的触发条件和控制方法，以及控制主权切换的介入时刻界定，最终实现人机驾驶切换的安全性和稳定性。基于以上研究工作，项目负责人在IEEE Intelligent Transportation Systems Magazine、Journal of Intelligent and Connected Vehicle、交通运输系统工程与信息等行业知名期刊和学术会议上发表了论文6篇，申请了发明专利5项，授权软件著作权4项；参与编制了国家标准“交通运输卫星导航增强服务性能指标及监测技术规范”和“交通运输卫星导航增强定位模块测试技术规范”；相关成果在昌九智慧高速公路（江西省新一代国家交通控制网和智慧公路示范工程）新祺周到永修收费站近10公里试验段开展了应用与测试；获得了第十二届全国交通运输领域青年学术会议优秀论文奖（排名第一）和2020年度中国交通运输协会科学技术奖一等奖（排名第三）；此外，通过本项目培养了硕士研究生3名。