impl RL for Nao

[oleflb]

We have made promising results with learning to Stand using RL. This may be a good idea to see everything come together and implement the learned policy on the Nao.

• Transfer neural network to openvino format
• Implement the get_obs function on the Nao using our sensor values
• Map the action array to a MotionCommand
• Add Safeguards? Maybe something like maximum velocity?
• Infer the policy on the nao during the Stand action
• Buy some 🍅 and test it :D