Open algo-scope opened 1 month ago
from my understanding, the system is run stage by stage, once a component fails, it could reverse to the last stage to retry.
As for openai connection, it will increase the overall time cost, but for each execution it will not involve. As IK is solved and operated directly by the pybullet in a more classical control way
Hello, thank you for your work! Due to my network conditions, there is approximately a 2-second latency each time I call OpenAI. I have 32GB of memory and a 12GB 3060 graphics card, but during the demo run, OmniGibson is a bit laggy. The saved video seems normal. I would like to know if this situation will affect the performance of your model, or will it only manifest as slower speed but with accurate actions?