Open CarbonGU opened 8 years ago
Can you explain a bit more what does it mean by "awkward post"? Do you have any screen capture? I never experience this problem (Other people also manage to install and run successfully)
On Mon, Oct 24, 2016 at 10:40 AM, Wen-Yi Gu notifications@github.com wrote:
Hi,
I try to running your code to train the torcs, but every time reset the torcs after 3 episodes, the initial screen input shows an awkward pose and I need to press F2 to fix it. Is any solution to it? Thanks
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/yanpanlau/DDPG-Keras-Torcs/issues/3, or mute the thread https://github.com/notifications/unsubscribe-auth/AO1sYzduMZSUewGZHvn_2JhSMUWBlPtmks5q3BqXgaJpZM4KeWBt .
Hi I mean I got a viewpoint which is not appropriate as a screen input to Neural Network. It's NOT the first-person driving view.
I had to press F2 to change the view, but when torcs was relaunched after 3 episodes , the driving view became non-first-person-view again.
I hope it looks like this:
But it looks like this after relaunch:
Do you have any ideas? thank you:)
Added these lines in the file autostart.sh may solve your problem
xte 'usleep 1000000' xte 'key F2' xte 'key 0'
I have the same questions my speed, driver board and other information all disappear but I have these problem at first time and can't be solved by pressing F2
@CarbonGU @Artioss @jiangdairong @yanpanlau Hi, have you solved the on_screen_information problem that no speed, driver board ? Thx!
@lkhit Hi, you can find the solution here https://github.com/ugo-nama-kun/gym_torcs/issues/7
@CarbonGU Have you solved the problem of shifting the viewpoint ?
Hi,
I try to use the raw screen image as input to train the torcs based on your code, but every 3 episodes restarting the torcs, the initial screen input shows an awkward pose and I need to press F2 to switch it.
Is any solution to it? Thanks
你好,我感觉你把当前视角图像作为神经网络的输入,是一件非常有意思的事情。但是,如何在torcs修改以得到屏幕截图?希望你能够提供帮助,谢谢!
Hi,
I try to use the raw screen image as input to train the torcs based on your code, but every 3 episodes restarting the torcs, the initial screen input shows an awkward pose and I need to press F2 to switch it.
Is any solution to it? Thanks