DBL: Driver Behaviour Learning

DBL：驾驶员行为学习

• 批准号: 519912-2017
• 负责人: Fischmeister, Sebastian
• 金额: $ 12.69万
• 依托单位: University of Waterloo
• 依托单位国家: 加拿大
• 项目类别: Collaborative Research and Development Grants
• 财政年份: 2019
• 资助国家: 加拿大
• 起止时间: 2019-01-01 至 2020-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=693997
• 关键词: DBL; Driver; Behaviour; Learning

The project goal is to develop machine learning technology to mimic driver behaviour under normal driving conditions in North America. Such learnt models would be applicable to improve the functionality of a variety of different features including, but not limited to Adaptive Cruise Control (ACC), Lane Change Assist, automated driving (SAE L1 to L4).Furthermore the project goal is to develop machine learning technology that supports updating while the driver is operating the vehicle. The application is to allow deploying basic models in vehicles at the point of sales, and then later have the vehicles adapt to driving behaviour of the owner.The project challenges are to identify base technology in machine learning suitable for the problem, then adapt it to the specific needs of the automotive industry and driving, and finally actually validating the technical feasibility, viability, and efficacy of the technology with real world data.

该项目的目标是开发机器学习技术，以模拟北美正常驾驶条件下的司机行为。这些学习的模型将适用于改进各种不同功能的功能，包括但不限于自适应巡航控制(ACC)、车道改变辅助、自动驾驶(SAE L1至L4)。此外，项目目标是开发支持在驾驶员操作车辆时进行更新的机器学习技术。该应用程序允许在销售点将基本模型部署在车辆上，然后让车辆适应车主的驾驶行为。项目挑战是确定适合问题的机器学习基础技术，然后使其适应汽车行业和驾驶的特定需求，最后使用真实世界的数据实际验证该技术的技术可行性、可行性和有效性。