For the next several weeks, TechCrunch’s robotics newsletter, Actuator, will be running Q&As with some of the top minds in robotics. Subscribe here for future updates.
Part 1: Matthew Johnson-Roberson of CMU
Part 2: Max Bajracharya and Russ Tedrake of the Toyota Research Institute
This time it’s Dhruv Batra, director of research, FAIR (The Fundamental AI Research) at Meta. Facebook’s parent describes FAIR as follows:
The Fundamental AI Research (FAIR) team at Meta seeks to advance our fundamental understanding in both new and existing areas, covering the full range of AI-related topics, with a mission to advance the state-of-the-art of AI through open research for the benefit of all.
Batra is also an associate professor at the Georgia Institute of Technology.
What role(s) will genetic artificial intelligence play in the future of robotics?
I see generative AI playing two distinct roles in embedded AI and robotics research:
- Data/experience generators
Create 2D images, videos, 3D or 4D scenes (3D + time) of simulated experiences (especially action/language preparation experiences) for robot training, because real-world experience is very rare in robotics. Basically, think of these as “learned simulators”. And I believe robotic research simply cannot be scaled without simulation training and testing. - Architectures for self-supervised learning
Generating sensory observations that an agent will observe in the future, to compare with real observations and use as an unannotated signal for learning. I see Jan’s paper at AMI for more details.
What are your thoughts on the humanoid form factor?
I’m optimistic about that. Essentially, human environments are designed for the humanoid form factor. If we really want general-purpose robots to function in environments designed for humans, the form factor should be at least somewhat humanoid (the robot will likely have more sensors than humans and may have more components).
How far are true general purpose robots?
Thirty years. So effectively outside the window where any meaningful prediction is possible. In fact, I think we should be deeply skeptical and suspicious of people who make “AGI is around the corner” claims.
Will home robots take off (beyond the gaps) in the next decade?
No, I don’t believe the core technology is ready.
What major robotics story/trend isn’t getting enough coverage?
That now we can test navigation robots in real houses and things really work! Notice, unlike self-driving cars, these home navigation robots don’t have the luxury of building accurate maps over millions of miles of driving. We take a robot to a new house and ask it to find objects.