End-to-End solutions for Multirobot Coordination

My vision is to create solutions for coordinating teams of robots that start from truly high level specifications and deliver code for individual robots in the system. Currently, multirobot systems are extremely difficult to use, and solutions are not transferrable from one application to another. The goal of my research is to develop distributed planning and control foundations that are broadly applicable across all aspects of multirobot systems or mobile sensor networks.

Social Computation

Teams of humans are exceptionally good at coordination. Teams of robots, however, are clumsy at coordination, requiring extensive communication and computation. As cognitive beings that make decisions based on broad context, memory, and sensing, human capabilities are challenging to transfer to robotics. However, by limiting context (communication and sensing), and actuation, we can constrain humans to robot-like capabilities. We are using online multiplayer games to provide a faithful representation of the capabilities of distributed teams of robots. Games are an ideal platform for studying how human capabilities can be applied to multirobot systems, since they provide full control over an interface where context (communication and sensing capabilities) can be easily added and removed for studying various scenarios. For more detail click here.

Mixed Reality for Robotics

Mixed reality is a valuable tool for research and development in robotics. When robots operate in shared environments with humans, they are expected to behave predictably, operate safely, and complete the task even with the uncertainty inherent with human interaction. Preparing such a system for deployment often requires testing the robots in an environment shared with humans in order to resolve any unanticipated robot behaviors or reactions, which could be potentially dangerous to the human. Our mixed reality approach fuses multiple physical and virtual worlds, allowing virtually unlimited combinations of physical and virtual agents. It can ease a system into sharing space with a human by, for example, running controllers with virtual humans, then humans in another physical environment, and only after those, in the same space. It also allows enhancing robots with virtual sensors, such as cameras, to do vision-based control or scaling up your multirobot system. Funded by ARL W911NF-14-D-0005. For more detail click here.

Resource Management for Long Term Depoloyments

While typically capable of short term autonomy, the vulnerabilities of exposed hardware and limited on-board power, combined with the shortcomings of current software and algorithms prevent long-duration autonomy of unmanned vehicles. Hardware that has been in use for extended time is subject to biofouling, misalignment, and breakdown. Energy can also be a significant limiting factor: it is well-known that the amount of power carried on-board smaller, more agile platforms severely limits the time a robot can be actively on mission before needing refueling. Increasing on-board power comes at the cost of reducing range or agility. We address the many challenges of long-term autonomy for multirobot systems including solutions for persistent control and coordination, and addressing resource degradation. Funded by the Office of Naval Research N00014-14-1-0734. For more detail click here.