车辆的智能化已成为世界公认的汽车行业发展趋势之一，精准的传感器与强大的计算决策能力可以弥补驾驶员操作的诸多缺陷。然而，面向车辆智能化应用的单一传感器由于易受环境因素干扰而无法实现可靠感知，现有多传感器融合算法大多采用前融合方式而无法有效结合异构信息各自的优势。在运动控制方面，现有方法大多完全依赖自动驾驶设备却忽略了驾驶员具有机器无法替代的优势。针对上述问题，本项目将开展三方面的研究：1、研究融合相机与激光雷达信息并结合时空连续信息来实现动态障碍物识别定位和道路的重构；2、研究对动态障碍进行运动预测的方法并基于预测结果进行动态轨迹规划；3、结合驾驶员的驾驶意图对预规划的轨迹进行局部优化并实现精确跟踪。本项目采用与传统人机切换方式不同的策略，将丰富车辆感知、规划与控制相关理论并推动智能车的商业化。

Intelligent vehicles have been recognized as one of the trends in the automotive industry. Precise sensors and powerful computing devices can help make decisions in case driver failures occur. However, when applied to intelligent vehicles, a single sensor is not reliable to perceive the environment as it is easy to be influenced by environmental disturbances. Existing multi-sensor fusion algorithms mainly adopt front fusion methods, and thus cannot take the advantages of heterogeneous multi-sensor information effectively. As for the vehicle motion control, most existing frameworks rely purely on the autonomous driving devices and ignore the irreplaceable strengths of human drivers. To deal with the problems mentioned above, this project will carry out research work in the following three aspects: 1. Recognition and localization of moving obstacles and reconstruction of drivable road by means of the fusion of cameras and LiDARs together with continuous spatial-temporal information. 2. Prediction of the movement of moving obstacles and dynamic trajectory planning based on the predicted information. 3. Local optimization of the pre-planned trajectory based on the driver intention and precise tracking of the optimized trajectory. Adopting a strategy different from the traditional man-machine switching methods, this project will enrich the perception, planning, and control theories related to intelligent vehicles and promote the commercialization of intelligent vehicles.