Shiwali Mohan
Principal AI Scientist at SRI International, Future Concepts (formerly Xerox PARC)
I build intelligent collaborative agents that model, learn, and reason about their human collaborators. My research advances methods for complex sequential decision making, intelligent agent systems, as well as hybrid AI agent architectures built with statistical machine learning and knowledge-rich inference algorithms. I leverage insights from economics, psychology, education, and HCI to desgin effective and robust human-agent collaboration in real-world settings.
As a Principal Investigator at SRI International, I lead technology and business development for AI research. My work spans fundamental algorithmic advances in agent architectures as well applications of agent technology in real world usecases.
Fundamental Agent Research: I lead research on open-world learning agents (DARPA SAIL-ON) and on teachable agents (DARPA GAILA). Both these efforts study how agents can adapt to new situations after they are operational without the need of taking them offline and re-training. I study the role of structured representations in efficient & resilient agent architectures and investigate how they can be manipulated or adapted efficiently on-the-fly.
Agent Applications: I have designed interactive, collaborative agents for a variety of domains including patient-centric, preventative healthcare, sustainable living, general purpose robots, and augmented reality. I am particularly motivated to build intelligent collaborative technology social good and public welfare.
My work is interdisciplinary and has been published at venues for research on artificial intelligence (AIJ, JAIR, AAAI, IAAI), human cognition (ICCM, ACS, BICA), human-machine interaction (ACM TiiS, IEEE RO-MAN) as well as in applications (JMIR, EMBC, ACM/AAAI AIES).
news
Sep 1, 2024 | Our work on open-world learning agents is published in the AI Journal. |
---|---|
Jul 9, 2024 | Patent on natural language interaction with robots is granted! |
Jun 4, 2024 | We demonstrated open-world learning for UAVs and LLM+planning for embodied agents at ICAPS 24. |
Sep 22, 2023 | Giving an invited talk at the Allen Institute of AI on advances in model-based reasoning systems. |
Jul 31, 2023 | I was invited to the DARPA AI Forward initative to identify the directions AI research should take next. |
selected publications
- ACSCharacterizing an Analogical Concept Memory for Architectures Implementing the Common Model of CognitionIn Proceedings of the Annual Conference on Advances in Cognitive Systems 2020
- AAAILearning Fast and Slow: Levels of Learning in General Autonomous Intelligent Agents.In Proeedings of the AAAI Conference on Artificial Intelligence 2018