search

merantix / imitation-learning

Autonomous driving: Tensorflow implementation of the paper "End-to-end Driving via Conditional Imitation Learning"
https://medium.com/merantix/journey-from-academic-paper-to-industry-usage-cf57fe598f31
MIT License
89 stars 21 forks source link

high level command may not very work #18

Open zhangjunwang opened 5 years ago

zhangjunwang commented 5 years ago

I have trained your implementation(29400 steps). And then i use this https://github.com/carla-simulator/imitation-learning to test in Carla. I set directions(high level command) as a constant,such as 4.0, it means the agent will turn right when it is in the intersection. However, it doesn't work. it can follow lane very well, but do not turn left or turn right following the command. Do you know why this happened? Very thanks

zhangjunwang commented 5 years ago

when i use this https://github.com/carla-simulator/imitation-learning trained model, It can follow the command.

Amakri1020 commented 5 years ago

I think the branch heads are not defined in the same order, you'll have to modify the run_CIL code to respond to the order used by this repo.

raks097 commented 4 years ago

@zhangjunwang Hi, Could you provide the implementation you used run the vehicle using the trained models inside the CARLA simulator.

Thanks