Collaborative Robots for Longer Autonomy with Enhanced Perception and Mobility
Abstract
Approved For Public ReleaseProblems concerned in this project are motivated by the limitations of the state-of-the-art robots in completing human tasks in real environments. Since robots should be used particularly for human safety, damage assessment and response in human-built complex environments are two of such tasks. They are also important for navy since vessels, providing human livingon a fuel storage farm and a power plant with high voltage, are subject to various dangers. This has given rise to need for robotic assessment and response, which allow robots to complete some of the tasks in collaboration with humans or in cooperation with the other robots. However, the fundamental problem of robotic assessment and response lies in the lack of robots in both actuation and sensing capabilities. While technical issues on mobility and perception have been tackled independently, they are closely correlated. The two problems should be considered together for longer autonomy. By leveraging the knowledge and facility developed in the past, the goal of this new project is to develop a team of heterogeneous robots that enable longer collaboration with humans or cooperation with the other robots in complex human-built environments. To achieve the goal, the technical objectives defined for the project are as follows: #Enhanced Mobility: Custom-design and develop robots that can move around complex human-built environments including the indoor environments of a navy vessel. Currently, there exists no robot that can move around such environments. The developed robots will additionally have mobility that enhances perception and achieves longer-term autonomy. #Enhanced Perception: Enhance perception by extracting and leveraging richer and more cognitive information. For collaboration with humans, the information will be human intention. For better cognition of structures, the project will focus on the assessment and response of damage. The enhanced perception proposed in this project will leverage the high-precision map developed in the preceding project. The developed robots will additionally have perception that minimizes robot motion. The proposed collaborative autonomy is characterized by the three unique in-house technologies. For enhanced mobility, the major robot development effort for complex human-centric environments has been exerted on humanoid robots which have excessive joint actuators to mimic humans. The proposed wheeled humanoid robots, meanwhile, are customized to achieve the targeted mobility with a minimal set of actuators while assisting enhanced perception for longer autonomy. The remaining two for enhanced mobility extract and leverage richer and more cognitive information. For collaboration with humans, the human intention is extracted and incorporated, which allows the longer prediction of human behavior more reliably and accurately. For the assessment and response of structural damage, the multimodal damage assessment and the damage response based on a new Learning from Demonstration (LfD) are proposed. The developed technologies will be ultimately evaluated and demonstratedin vessel-like environments. Tasks to demonstrate are all for human collaboration and damage assessment and response. Detecting damage and conducting position-controllable tasks in the vessel-like environments semi-autonomously in collaboration with humans will be first challenged. In hardware, mechanical capability of the developed robots in mobility and manipulation will be investigated and validated. In software, collaborative autonomy of the developed robot, with a human and with other robots, will be tested and evaluated.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jun 29, 2023
- Source ID
- N000142312572
Entities
People
- Tomonari Furukawa
Organizations
- Office of Naval Research
- United States Navy
- University of Virginia