个性化人车交互及智能共驾系统的研究与实现

Driving is a complex, dynamic and time-varying process. Different drivers vary in driving habits and operation style. This project focuses on exploring the way emotion, physical fatigue and driving habits influences a driver’s behaviour behind the wheel by investigating the driver-vehicle cooperation protocol under the framework of human-machine interaction. In order to enhance the mutual communication and to encourage cooperation between a driver and the vehicle, we are motivated to build a smart driving assistance system which owns self-learning capability through sensing, signal processing techniques and intelligent learning methodology. By inspecting the spoken dialogues from the in-car speech interaction architecture and by analysing the driver’s behaviour, we can build a driver-specific model, which includes his/her interaction modality, emotional burden, muscle fatigue and operation habits, etc, for driver intent prediction and abnormality detection purposes. By integrating human-vehicle interaction platform, driver model and driving-assistance system together, and through exploitation of the human-machine cooperation theory and relevant strategies under driving environments, we aim to achieve collaboration between driver and machine under various driving conditions, especially when abnormality and fatigue occurs. The proposed study endeavours to improve the personalization and intelligence of smart cars, and to offer solutions for real-world driver-vehicle cooperation missions.

驾驶过程是一个复杂多变的动态系统，不同的驾驶员拥有自己的驾驶习惯和操作模式。针对驾驶员的情绪变化、生理疲劳和操作习惯对驾驶行为的影响，本项目采用状态感知、信号处理与智能学习的方法，在人机交互的框架下开展个性化人车协作驾驶的研究，致力于建立一个能够自主学习的智能驾驶辅助平台，以提升驾驶员与汽车之间的双向交流与协调合作。通过对人车交互过程中的语音对话和驾驶行为进行分析，获取驾驶员身份信息，并建立包括交互模式、情绪负载、肌肉疲劳和驾驶习惯在内的个性化驾驶员模型，进而预测驾驶员的驾驶意图并监测异常驾驶行为。本项目拟结合人机交互平台、驾驶员模型和辅助驾驶系统这三个智能驾驶的核心要素，通过对人机交互机理与人机共驾机制的深入研究，提出在不同场景下，尤其是异常或疲劳驾驶的情况下，实现人车共同驾驶的解决方案。上述研究是智能汽车向人性化与个性化发展的关键技术途径，对于促进智能汽车驾驶技术的发展具有重要意义。

驾驶过程是一个复杂多变的动态系统，不同的驾驶员拥有自己的驾驶习惯和操作模式。针对驾驶员的情绪变化、生理疲劳和操作习惯对驾驶行为的影响，本项目采用状态感知、信号处理与智能学习的方法，在人机交互的框架下开展了个性化人车协作驾驶的研究，建立了一个能够自主学习的智能驾驶辅助平台，以提升驾驶员与汽车之间的双向交流与协调合作。通过对人车交互过程中的语音对话和驾驶行为进行分析，获取驾驶员身份信息，并建立了包括交互模式、情绪负载、肌肉疲劳和驾驶习惯在内的个性化驾驶员模型，进而预测驾驶员的驾驶意图并监测异常驾驶行为。本项目结合了人机交互平台、驾驶员模型和辅助驾驶系统这三个智能驾驶的核心要素，通过对人机交互机理与人机共驾机制的深入研究，提出在不同场景下，尤其是异常或疲劳驾驶的情况下，实现人车共同驾驶的解决方案。上述研究是智能汽车向人性化与个性化发展的关键技术途径，对于促进智能汽车驾驶技术的发展具有重要意义。