Embodied Intelligence
AAAI 2023 Inaugural Summer Symposium Series
AI has recently begun to pay more attention to embodied intelligence. In psychology, the idea of embodied cognition stipulates that the high-level mental processes, including conceptualization and reasoning, are intimately linked with the seemingly “low-level” of processing in an intelligent entity (human or animal) involving the motor and sensory systems. Some AI research in recent years has begun to direct its attention onto this aspect of intelligent functioning, and the field has seen an increasing number of papers and workshops. In this workshop we propose to specifically look at aspects of embodied intelligence that involve the interactions between humans and robots – human-human, human-robot, and robot-robot. Report on research on all other aspects of embodied intelligence are welcome.
We welcome submissions in two formats: (a) short papers (2-4 pages) and (b) full papers (6-8 pages). Papers should be formatted using the standard AAAI template. Submissions should be done through the AAAI EasyChair Submission site.
Selection of submissions follows a juried process where authors may expect to receive light feedback. Accepted submissions will be published in AAAI symposia proceedings for authors who wish to be included.
Submissions due: June 30, 2023
Author Notification: within 10 days of submission
Topics
Papers related to, but not limited to, the following issues are welcome:
How considerations of embodied intelligence impact the representational and computational issues of AI agents
Scene graphs and other representational structures appropriate for capturing knowledge for embodied agents
Visual attention issues on embodied agents
What information is important for human-robot interactions, inspired by human-human interactions
Appropriate behaviors of robots in human social environment
Recognizing human intentions through video and other cues
Symposium Schedule (18 July)
9.00am to 9.30am: Some of the many unexplored problems in embodied intelligence, Andrei Barbu (MIT)
9.30am to 10.00am:Visual Intelligence from Human Interactions. Ranjay Krishna (UW)
10:00am to 10:30am: Democratizing Robot Learning for All, Jiafei Duan (UW)
11:00am to 12:30am: Spotlight paper talks
Neural Amortized Inference for Nested Multi-agent Reasoning
(Kunal Jha, Tuan Anh Le, Chuanyang Jin, Yen Lin Kuo, Joshua Tenenbaum,and Tianmin Shu)
Small Object Navigation with Context Information
(Jiaming Wang and Harold Soh)
Integrating Curricula with Replays: Its Effects on Continual Learning
(Ren Jie Tee and Mengmi Zhang)
12:30am to 2:00pm: Lunch
2:00pm to 2:30pm: Embodied AI in Body's Eye ,Yixin Zhu (Peking University)
2:30pm to 3:00pm: Teaching robots to infer futures for robust human-AI interactions, Yen-Ling Kuo (University of Virginia)
3:00pm to 3:30pm: The Path to AGI Goes Through Embodiment, Cheston Tan (A*STAR)
3:30pm to 4:00pm: Tea Break
4:00pm to 4:30pm: Do Vision-Language Models Have Socio-Physical Understanding? Rawal Ishaan Singh (A*STAR)
4:30pm to 5:00pm: Cracking the Context Code: Advancing Vision with Context-aware Modeling, Mengmi Zhang (A*STAR & NTU)
Organising Committee
Center for Frontier AI Research (CFAR), A*STAR
Institute of High Performance Computing, A*STAR
Center for Frontier AI Research (CFAR), A*STAR & Nanyang Technological University of Singapore
University of Washington, USA
University of Virginia
Singapore Management University
Singapore Management University