Rhoda AI: 1000X less training data required?

<p>Can robots learn from the internet the same way ChatGPT learned from text?</p><p><br></p><p>In this episode, Andrew Wooten, co-founder of Rhoda AI, explains why his company believes the future of robotics isn’t collecting millions of hours of robot data ... it’s learning from internet-scale video. Instead of relying on traditional vision-language-action (VLA) models that require enormous training datasets, Rhoda’s approach teaches robots physical intuition by predicting the future through video.</p><p><br></p><p>We also explore why, in Andrew&#39;s opinion, warehouses and factories will likely be the first major market for humanoid robots (not homes!), why Rhoda chose a wheel-based humanoid design, how language models fit into physical AI, and how the company’s robots can learn complex tasks with just 8–10 hours of training data instead of 10,000+ hours.</p><p><br></p><p>If you’re interested in robotics, AI, automation, or the future of manufacturing, this conversation offers a fascinating look at where physical AI is heading.</p><p><br></p><p>In this episode:</p><p><br></p><p>* Why warehouses beat homes as the first market for humanoid robots</p><p>* Why Rhoda chose wheels instead of legs</p><p>* The biggest limitation of today’s robot AI models</p><p>* How internet-scale video teaches robots physics</p><p>* Why predicting the future helps robots manipulate the real world</p><p>* Edge AI vs. cloud robotics</p><p>* The role of LLMs in controlling robots</p><p>* How Rhoda cut robot training from 10,000+ hours to just 8–10 hours</p><p>* When zero-shot robot learning could become reality</p><p><br></p><p>Guest</p><p><br></p><p>Andrew Wooten</p><p>Co-founder, Rhoda AI</p><p>Website: https://rhoda.ai</p><p><br></p><p>00:00 Why Humanoid Robots Don’t Have Wheels</p><p>00:18 Can Robots Learn From Internet Video?</p><p>00:42 Best Use Cases for Humanoid Robots</p><p>02:10 Why Warehouses and Factories Come First</p><p>04:02 The Economic Impact of Robotics</p><p>05:00 Home Robots vs. Industrial Robots</p><p>06:05 Why Rhoda AI Chose Wheels</p><p>08:00 Building a General-Purpose Robot</p><p>09:55 Why Full-Stack Robotics Companies Have an Advantage</p><p>10:40 The Evolution of Physical AI</p><p>12:20 Why Vision-Language-Action Models Fall Short</p><p>14:05 Training Robots With Internet-Scale Video</p><p>16:05 How Rhoda AI’s Video-Action Model Works</p><p>17:25 Edge AI vs. Cloud Computing</p><p>19:05 How Robots Develop Physical Intuition</p><p>20:40 Predicting the Near Future in Real Time</p><p>22:00 Can Robots Build a Subconscious?</p><p>23:10 Using Language Models to Control Robots</p><p>25:15 Rhoda AI’s Hardware Strategy</p><p>26:20 The Biggest Problems With Today’s Humanoids</p><p>28:20 When Will Robots Truly Learn on the Job?</p><p>30:05 Training Complex Tasks in 8–10 Hours</p><p>31:15 Zero-Shot Robot Learning and What Comes Next</p>