LEADERBOARD

DISCRETE
CONTINUOUS
OBSTACLES, DISCRETE
OBSTACLES, CONTINUOUS

The best result in OffWorldMonolithDiscreteReal-v0 environment is 0.96 by Ashish Kumar!
Run your policy in evalutaion mode by calling  gym.make(..., mode='test')  to see it here.

Experiments conducted: 228
Episodes finished: 68597
Hours of real learning logged: 1715

TOP 3 IN END TO END


sac_monolith_discrete_real_12_12_16AM_Jun-26-2020
by JB (offworld), average test reward 0.93

ddqn_e2e_experiment_3
by Ashish Kumar, average test reward 0.92

DQN, depth 320x240, win 1, ann 40k
by Ashish Kumar, average test reward 0.44

TOP 3 IN SIM TO REAL


TOP 3 IN HUMAN DEMONSTRATIONS


dqn_depth_hl_experiment_5
by Ashish Kumar, average test reward 0.96