Kaushik Balakrishnan, PhD

Hi, I'm having the same issue. Did you solve the problem?

Actually I solved the problem. We need to also install vtorcs-RL-color. The relaunching loop issue is solved for me!

But I am trying to train from scratch. I get a frozen screen and the message "Timeout for client answer." Were you able to train from scratch? Also, the weights...

I suppose you are referring to line 67 of this file: vtorcs-RL-color/src/drivers/scr_server/scr_server.cpp which currently is set to: #define UDP_DEFAULT_TIMEOUT 100000 Should I increase this number?

Works now. That's in microsecs actually. Thanks a lot!

One final question. To train from scratch, how many episodes did you need for the car to drive properly? It is currently set to the following values: episode_count = 2000...

Thanks, I got it working after some minor changes.

Somewhere near line 67 of vtorcs-RL-color/src/drivers/scr_server/scr_server.cpp you can set: #define UDP_DEFAULT_TIMEOUT 100000000 (this number is time in microseconds, set to a large value) After that, the following 3 commands to...

Yes, I observed this too. Line 54 of the critic should be: V = Dense(1,activation='linear')(h3) instead of V = Dense(action_dim,activation='linear')(h3) Note that in the Bellman equation we have r +...

Type the following command in another terminal when you train the model: python -m visdom.server