CAREER: Towards Autonomous Wheelchair Systems in Urban Environments

“这项奖励是根据2009年美国复苏和再投资法案（公法111-5）资助的。”预计美国人口老龄化将给这个国家带来前所未有的压力。美国的医疗保健系统。为了缓解这种情况，越来越多的行业对以家庭为中心的医疗保健方法感兴趣。利用技术进步被视为家庭护理范式的主要推动者之一。为此，该项目正在研究智能轮椅技术的发展。有三个重要的技术组成部分。第一个是使用大规模的三维地图，即使在没有可靠的GPS测量的情况下，也可以在室外和城市环境中支持轮椅的精确定位。这是通过自动合成先验的地图模型来定位轮椅在运行时可以使用的稳定、稳健的地标特征来完成的。其动机是，通过利用高保真/成本传感器系统生成的地图，低成本服务机器人可以获得这些传感器的许多好处，但无需成本。这对该技术未来的商业可行性至关重要。第二是基于优化的人群运动规划框架，使智能轮椅能够在大量行人和静态/动态障碍物存在的情况下安全机动。该方法基于凸优化理论的最新进展，并通过交互控制功能明确地集成了轮椅-行人动力学的高级模型，以提高导航性能。最后，通过与宾夕法尼亚州Allentown的Good Shepherd康复医院合作，计划进行重要的开发和测试。这将为系统开发提供关键的服务提供者和最终用户输入。

"This award is funded under the American Recovery and Reinvestment Act of 2009 (Public Law 111-5)."The aging U.S. population is expected to place unprecedented strains on the country?s health care system. To mitigate this, there is growing industry interest towards home-centered approaches to health care. Leveraging technology advances is seen as one of the primary enablers of the home-care paradigm. To this end, this project is investigating the development of smart wheelchair technologies. There are three significant technical components. The first isthe use of large-scale, three-dimensional maps to support accurate wheelchair localization in outdoor, urban environments even in the absence of reliable GPS measurements. This is accomplished through automated synthesis of map models a priori to locate stable, robust landmark features that the wheelchair can use at runtime. The motivation is that by leveraging maps generated from high fidelity/cost sensor systems, lower-cost service robots gain many of the benefits of these sensors but without the cost. This is critical for the technology to be commercially viable in the future. The second is an optimization-based framework for motion planning in crowds that will enable the smart wheelchair to maneuver safely in the presence of large numbers of pedestrians and static/dynamic obstacles. This approach builds upon recent advances in convex optimization theory, and explicitly integrates higher-level models of wheelchair-pedestrian dynamics through interactive control functions to improve navigation performance. Finally, there is a significant development and test component planned through a collaboration with Good Shepherd Rehabilitation Hospital in Allentown, PA. This will provide critical service provider and end user input into system development.