NRI: Decision-making Foundations for Human-supervised Legged Robot Teams

点击翻译按钮获取中文摘要

Mobile robots are emerging as a powerful tool in many applications that require highly repetitive, tedious work or pose a threat to human health. Wheeled robots are adequate for many applications in structured environments such as hospitals, warehouses, and grocery stores. However, many other applications require robots to traverse challenging terrain. Legged robots will be more suitable for applications such as disinfection of cluttered spaces, pesticide spray, and inspection of construction sites. They can navigate around and over obstacles, climb stairs and slopes, or traverse muddy terrains to perform functional tasks. In many critical missions, human-robot teaming is emerging as an attractive option. Robots can keep humans away from physical dangers, and humans can make high-level decisions to ensure robots are taking the right actions based on the mission's needs. Therefore, this research project will study decision-making foundations that will enable a team of legged robots to work safely and effectively with human supervisors on real-world applications. The project will also integrate research results into robotics and control courses. Outreach activities will educate and inform K-12 students about career opportunities in robotics.Legged robots navigating challenging terrains while performing critical tasks are expected to encounter many challenges. Terrain maps are not likely to be completely known in advance. Moreover, there might be moving obstacles in the work environment. Robots will have to update maps and navigate around moving obstacles as a part of the task execution. On-board sensors are likely to experience occlusion and produce a significant amount of uncertainty. Therefore, robots will need to account for the uncertainty in risk assessment. Safety is also a crucial consideration. Legged robots will need to ensure that they can walk safely. However, efficiency in execution will be an essential consideration for deploying legged robots in real-world applications. To achieve task execution efficiency, robots will need to carry out functional tasks while moving. This project proposes to develop (1) methodologies for human-supervised robot team mission modeling and task assignment by considering risk and mission execution efficiency, (2) develop algorithms for risk-aware deliberative motion planning by considering future contingency situations and actions needed to resolve them, and (3) multi-objective safety-critical controllers for mitigating risk, ensuring stability and safety for both locomotion and task execution.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.