AI-assisted Dynamic Adaptive Planning for Human-in-the-Loop Multi-Agent Systems

Principal Investigator: Shaoshuai Mou

Topic 3 “Quantitative Dynamic Adaptive Planning” in Northrop Grumman's Research in Applications for Learning Machine Consortium (REALM). We aim to develop a AI-assisted Multi-Agent platform, which

• is able to provide distributed environment perception/situation awareness (on-board distributed fusion, AI-assisted object recognition)
• is able to perform real-time, dynamic and distributed control/management of assets, planning/decision making/task assignment.
• is able to integrate human inputs in natural language/human gestures and provide feedback/suggestions to human commander. (AI-assisted natural language processing, imitation learning, mixed-human-robot autonomy)
• is able to improve performance as time evolves (lifelong learning).

The platform will target at applications information gathering, search and rescue in response to nature disaster response. 

The proposed research will be performed by a collaboration of four faculties from three different university with diverse research backgrounds. The team as a cohesive whole works together with Neta Ezer, Hasan A. Ghadialy and other Technical Leads from Northrop Grumman.

Personnel

Other PIs: Dan Delaurentis, Xinhua Zhang, Joydeep Biswas

Students: Xuan Wang, Wanxin Jin, Paulo Heredia, Nick Schultz

Keywords: AI, Autonomy, Distributed, Dynamic Environments, Learning, Multi-Agent Networks, Real-Time Management