Q - Where are we in the state of humanoid robotics? -
A - We are I think we’re right on the brink, I would say I would say we went past the point of technical capability definitively last March (2023). From that point onwards, there were no there were no fundamental unknowns anymore. For a long time. There's always been something we didn't know how to do. - Rob Knight
Q - What was the last to fall? Jeff Frick
A - I would say it was when the LLMs were shown that they could generate chains of motor commands to solve complex tasks automatically. And the company that demonstrated it was Agility [Robotics]. - Rob Knight
'Clean up this mess.'
What is a mess?
What is 'clean up'?
What can I successfully engage with?
What does successful engagement, in context of the task, look like?
LLMs, LVMs (large visual models), Reinforcement learning, teleoperations learing, ChatGPT, and more AI based technologies are to the point they can successfully hear, understand, and covert speech, to intent, an operational plan, robot movement control language, and execution of tasks.
The breakthrough demo was almost 2 years ago.
Using an LLM to direct our robot Digit. #AI #GPT - Agility Robotics YouTube Channel, 2023-May-30
Where do you think we'll be 2 years from today?
Really interesting conversation with one of the Open Source Hand gurus in robotics, Rob Knight, The Robot Studio
Rob's Mission: Democratizing Robotics Through Open Source | Humanoids Summit YouTube
#Robots #Robotics #Humanoid #HumanoidRobotics #LVM #Automation #TechnicalBreakthrough #LLM #AgilityRobotics #AI #GenAI #Labor #Future #Technology #MachineLearning #RoboticsTechnology #FutureOfAI #Innovations #Interview #HumanoidsSummit
Share this post