Physical Reasoning in Artificial Agents
The bulk of my research focuses on reverse engineering human physical reasoning to understand how we conceptualize and interact with the world. However, the same principles that are used to understand human reasoning can also help to design more human-like AI systems. Across a set of projects, I have helped to translate findings from cognitive science to build artificial systems that can flexibly exist and act in the physical world.
Related publications:
- Modeling expectation violation in intuitive physics with coarse probabilistic object representations
- Physion: Evaluating physical prediction from vision in humans and machines
- Physion++: Evaluating physical scene understanding that requires online inference of different physical properties
- Data-efficient learning for complex and real-time physical problem solving using augmented simulation
- Are deep neural networks SMARTer than second graders?
- H-SAUR: Hypothesize, Simulate, Act, Update, and Repeat for understanding object articulations from interactions
- Unsupervised discovery of 3D physical objects from video
- End-to-end differentiable physics for learning and control
- Differentiable physics and stable modes for tool-use and manipulation planning
