Chelsea Finn, co-founder of Physical Intelligence and Stanford associate professor, discusses building general-purpose robots with the No Priors hosts.
Finn traces her interest in robotics to its potential impact and the challenge of developing machine perception and intelligence. Her early work trained neural networks to map images directly to robot arm movements, but she found that getting a skill to work across many objects and settings remained difficult.
Physical Intelligence aims to build a neural network that can control different robots across varied tasks and situations. Finn says data from different robot designs can transfer, helping preserve what a model has learned when a company changes its hardware. The team uses Transformers and pretrained vision-language models alongside data collected by teleoperating real robots.
Finn identifies more diverse robot data as the main near-term need, including data from more places, objects, and tasks. She says internet video and pretrained models can help, but robots also need experience from their own bodies to learn motor skills. Deployment is difficult because small contact errors can prevent a task from succeeding, and many robot actions lack a human checking each result.
Finn describes a hierarchical system in which one model interprets a request and chooses the next step, while another turns that step into motor commands. The team demonstrated tasks such as making vegetarian and ham-and-cheese sandwiches, grocery shopping, and cleaning tables. Finn says current policies rely on camera images and lack memory, and she prioritizes adding memory over new sensors; she also favors robot designs that make data collection easier.
Finn says robotics may still be early, while smaller companies can move quickly to collect data outside controlled campuses. Her startup advice is to deploy, learn, and iterate quickly. Looking ahead, she expects many robot designs optimized for different jobs, rather than one universal form.
Briefs summarise talks and interviews; timestamps open the video at that moment. Claims in a talk are the speaker's own.
Chelsea Finn of Physical Intelligence discusses building reliable, general-purpose robots, drawing on office coffee service, factory box assembly, and laundry-folding examples.
Chelsea Finn, cofounder of Physical Intelligence, presents at a Y Combinator event on building reliable, general-purpose robots for real-world work.
Chelsea Finn, co-founder of Physical Intelligence, discusses building general-purpose robots at a Y Combinator event.
Stanford AI and robotics professor Chelsea Finn talks with host Peter on The Robot Brains Podcast about helping AI systems adapt to changing conditions, with examples from language models and robots.
Chelsea Finn, a Stanford professor, discusses how robots can generalize through broader training data and adapt to unfamiliar settings at a CMU Robotics Institute seminar.
Stanford University and Google Brain Professor Chelsea Finn presents her RI Seminar at Carnegie Mellon Robotics Institute on scaling data and algorithms for robot learning, introduced by host Wen.
