#Enjoy python enjoy.py --algo sac --env RoboSkateNumerical-v0 --folder ../trained_models/RoboSkate/RoboSkateNumerical-v0/ --load-best --env-kwargs headlessMode:False random_start_level:False startLevel:0 --exp-id 2 --n-timesteps 2000

sac/RoboSkateNumerical-v0_1

Was only trained to level 2. Can solve level 3 without having seen it.

sac/RoboSkateNumerical-v0_2

sac/RoboSkateNumerical-v0_1 was used as a pre-trained model and it was always started in level 1 and trained through all levels.

sac/RoboSkateNumerical-v0_3

Was trained from scratch with max_Joint_force = 350.0 without elif abs(board_roll-0.5) > 0.35: # Stop if board is tipped self.reward -= 10 print("board tipped") done = True elif abs(self.state.boardCraneJointAngles[3] * max_Joint_vel) > 200: # Stop if turning the first joint to fast "Helicopter" self.reward -= 10 print("Helicopter") done = True

and only on level 1

sac/RoboSkateNumerical-v0_4

Was trained from scratch
use agent_without_limits = True
with max_Joint_force = 1000.0
without
elif abs(board_roll-0.5) > 0.35: # Stop if board is tipped self.reward -= 10 print("board tipped") done = True elif abs(self.state.boardCraneJointAngles[3] * max_Joint_vel) > 200: # Stop if turning the first joint to fast "Helicopter" self.reward -= 10 print("Helicopter") done = True

and only on level 1

sac/RoboSkateSegmentation-v0_1

Trained on level 1 with image data latent space of VAE size 2

sac/RoboSkateSegmentation-v0_2

Trained on level 2 with image data latent space of VAE size 2

sac/RoboSkateSegmentation-v0_3

Trained on level 3 with image data latent space of VAE size 2