How to test my own RL algorithm in dukcietown with the graphics.

[zhaoworking]

I have suceessfully passed the Test "./manual_control.py --env-name Duckietown-udem1-v0". But I wanna test my own RL algorithm. So can u tell me how to do that? Should i creat a file first? and run it in the simulator. And how should i make a virtual display in which i can observe how the car runs.In other words, i wanna test my RL algorithm wit the graphics in the duckuietown.

[zhaoworking]

please help me , i get stuck in it.

[liampaull]

Hi - can you take at look at the AIDO RL baseline. The vanilla RL one with DDPG is here, and there's a new residual policy learning baseline (beta) here. Let us know if those are useful.

[zhaoworking]

Hi - can you take at look at the AIDO RL baseline. The vanilla RL one with DDPG is here, and there's a new residual policy learning baseline (beta) here. Let us know if those are useful.

thank u, but why can't i see my simulator when i am testing my RL with DDPG. How do i display my simulator when i test it.

[zhaoworking]

Hi - can you take at look at the AIDO RL baseline. The vanilla RL one with DDPG is here, and there's a new residual policy learning baseline (beta) here. Let us know if those are useful.

I have found the file called 'manual_control.py', and i can see the simulator and map when i was testing .but when i test the initial file 'train_reinforcement.py' , i can't see that.