MRI Track 1: Acquisition of a Platform for Education and Research in Risk Aware Robotic Exploration of Unknown Hazardous Environments

University Of New Haven, West Haven CT

Investigators

Abstract

This Major Research Instrumentation (MRI) grant is for the acquisition of a legged robotic platform at University of New Haven. The research topics to be pursued using this platform are aimed towards automated search and rescue operations in settings that may be hazardous for humans. This will help develop new knowledge related to risk-aware automated exploration of unknown hazardous environments, which promotes the progress of science, helps advance preservation of national health and helps with national welfare during disaster scenarios. The task of letting a robot autonomously navigate and explore an unknown environment is challenging. To handle this challenge, previous efforts have directed the robot to acquire a sense of where it is, where other objects are, sometimes trying to identify what these other objects around it are, and assess if such objects are moving and may collide with the robot. While such an approach may work in certain settings, it does not work in all, especially not when aiding post-disaster search and rescue operations, because a robot may not have enough abilities to understand all information its sensors receive. Current results in robotics lacks a detailed understanding of the consequences that the action of a robot can have on the overall search and rescue operation, on other humans who are supporting such an operation, or on those needing help in such a situation. The equipment supported by this award will promote multidisciplinary research related to artificial intelligence, autonomous control, electrical and mechanical engineering. It will also aid in educating the next generation on advancement robotic equipment and techniques. Such talent development initiatives have the potential to enhance national interests by developing a workforce trained to create, handle, and maintain the next generation of sophisticated artificial intelligence enabled robotics technology. The equipment acquired with this grant will help developing fundamental results related to collaborative risk-aware human-robot operations in an ever-changing, non-stationary disaster-stricken scenario. The following techniques are planned: theoretical guarantees from universal adaptive stabilization for near real-time in-scene object dynamics estimation; natural language processing based action sequence generation incorporating the evolving temporal status of task execution; enabling agile coordination and situational awareness in heterogeneous multi-robot settings via communicating multi-agent reinforcement learning algorithms; uncertainty quantification and risk-aware learning for safe and robust inter-robot communications via theory of mind and adversarial regularization approaches. 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.