Is LLMs for embodied control a great and right direction? or there are other better ways?
I think it is very limited since it's a closed loop and only capable of understanding high-level concepts.
Kevin Barrera Llanga
Dear Yusuke, personally, LLMs still have a viable field of application in robotics and integrated control, despite their limitations. They stand out for their efficiency, real-time performance, integration with hardware and sensors, and low resource consumption. However, for complex tasks that require adaptability, AI and machine learning may be more suitable. Combining LLM for specific control tasks with AI for decision making may be a strategy that expands the field of LLM.
0 votes
0 thanks
- Badges
- Science topic
More Yusuke Mikami's questions See All
Similar questions and discussions