Closed qgallouedec closed 6 months ago
For now I've dropped the use of the reward module, but can still add it back later if we want a support for diverse reward functions (not sure if it's useful though)
I took the idea of a dedicated reward library in manipulation from the meta-world paper and framework (https://meta-world.github.io/) - I will make a post on the slack on that subject - let's make it as you suggest for the moment.
For now I've dropped the use of the reward module, but can still add it back later if we want a support for diverse reward functions (not sure if it's useful though)