Autonomy of Service Robots in Cluttered and Dynamic Environments

前沿机器人学研究的重点是服务机器人和无人驾驶车辆在动态环境中的自主性，在动态环境中，其他物体可以移动和杂乱在任务空间中。今天的机器人被期望在静态和结构化的工作空间中做更多的重复性任务。他们必须执行复杂的行动，从战斗到堵住泄漏的海上石油管道，到移除脑瘤，到警告司机，再到吸尘繁忙的走廊。为了消除人为错误并使机器人在这样的现实世界应用中完全自动化，人们必须处理以变化的边界以及可移动、可变形和铰接式周围物体为特征的动态环境。这些看似不同的应用的共同需求是机器人感知环境并定位自己的能力，即感知，规划适当的动作，即运动规划，并最终执行规划，即控制。在动态环境的背景下，特别的挑战是这些步骤是耦合的，并且必须同时执行，因为例如，由于移动的障碍物导致的环境的微小变化将使先验计划过时。捕捉动态环境的动机不仅与提高自主性的愿望有关，而且还与新兴机器人应用和使用无人驾驶地面、航空和水下航行器所需的安全性和精确度有关，在这些地方，人工干预代价高昂、有限或不可能。这项研究致力于在动态环境中为传感、规划和控制提供健壮且计算高效的解决方案，以开发下一代自主机器人的具体细节，以应对现实世界应用的挑战。申请者有经过证明的记录，有能力和上进心的团队和最先进的实验室，能够成功地进行拟议的研究。研究人员对这一领域的兴趣日益增长，主要是因为一个机器人或一组协调工作的机器人在动态环境中工作的能力将开辟许多新的研究前沿和应用，从军事系统到交通安全再到环境研究，并对我们的生活质量、健康和经济产生巨大影响。

Leading-edge robotics research focuses on autonomy of service robots and unmanned vehicles in dynamic environments where other objects can move and clutter the task space. Today's robots are expected to do more than repetitive tasks in a static and structured workspace. They must carry out complex operations ranging from fighting battles to plugging leaky offshore oil pipes to removing brain tumors to warning drivers to vacuuming busy corridors. To eliminate human error and fully automate a robot for such real-world applications, one has to deal with dynamic environments characterized by changing boundaries as well as movable, deformable and articulated surrounding objects. The common need of these seemingly different applications is the ability of robots to perceive their environments and locate themselves i.e., sensing, plan an appropriate action i.e., motion planning, and finally execute the plan i.e., control. The particular challenge in the context of dynamic environments is that these steps are coupled and have to be executed concurrently since, for example, a slight change in the environment due to a moving obstacle will make an a priori plan obsolete. The motivation of capturing dynamic environments is related to not only the desire for an increased level of autonomy but also safety and precision required for emerging robotics applications and the use of unmanned ground, aerial and underwater vehicles where human intervention is costly, limited or impossible. This research focuses on robust and computationally efficient solutions for sensing, planning and control in dynamic environments to develop the nuts and bolts of the next generation of autonomous robots to cope the challenges of real-world applications. The applicant has a proven record, capable and motivated team and state-of-the-art laboratory to succeed in the proposed research. Interests in this field are increasingly growing among researchers mainly because the ability of a robot or a coordinated group of robots to work in a dynamic environment will open numerous new research frontiers and applications ranging from military systems to transportation safety to environmental studies, and create a great impact on our quality of life, health and economy.