Nov 14 2022 10:34 PM
I'm training humanoid to walk, stand on one leg, and so on in Bonsai. But I always start training from initial random policy so the brain has to learn from keeping balance to avoid falling.
It would be more efficient to train one brain to maintain a standing posture and then use that as an initial strategy to train applied movements above.
To do so, is it possible to transfer a learned model as an initial policy in another training?
Jan 10 2023 01:41 PM
Jan 10 2023 08:36 PM
SolutionJan 10 2023 11:47 PM