DeepDriving
DeepDriving copied to clipboard
Memory sharing problem
Hi
I followed the instructions from the original website to test the automate driving on three-lane. I start the torcs first, and then the Caffe. But it seems the Caffe can't get data from torcs. I noticed the two Memory sharing address for torcs and Caffe are different. Could I ask if you had met this problem before? I use tow terminals to open torcs and Caffe separately.
I didnt face any such problem. Btw did u follow the instructions exactly as in the website? You dont have to install caffe seperately.
I didn't install a new caffe, I just use the caffe provided by the author and make it.
I am not sure then. I followed the instruction in the website and it worked fine without issues.
You can also try installing gym-torcs from here https://github.com/ugo-nama-kun/gym_torcs. It has good python interface. I am currently working on this. This is much easier.
I have same problem as peterxu's one. I tried exactly same steps in deep driving read me.
- Install all the dependency of TORCS,
- cd to the "torcs-1.3.6" folder provided, type the following commands to install TORCS: ######################
- set chown on torcs
- edit .bashrc ######################
- Copy the modified tracks files in the "modified_tracks" folder to the torcs folder (/usr/local/share/games/torcs/tracks/road)
- Install prerequisites of Caffe
- Make Caffe in the provided folder "Caffe_driving" by typing "make"
and Then, run torcs first and then caffe. I tried data collection first, but I cannot get any data from torcs. And then, I tried torcs_test.sh with pretrained data with setting in README.
Is there any other settings needed for torcs which not in mentioned in README? I checked torcs tutorial but I cannot find additional settings for memory share.
Hi, have you solved your problem?
I have run it without any problem, and I find that the memory address for TORCS and Caffe are really different, maybe this is not your real problem. If you see three windows without any images after run torcs_test.sh, that's just ok.You just need to press P in a visulize window to start caffe, and then press C, the car will move. If there are not any windows shown on the screen, you may need to see the terminal window for error messages.
Could you please tell me where the instructions are?
Sorry. I didn't figure out the problem and how to solve. I'm sorry I can not help you.
2017-09-30 12:18 GMT+09:00 larenzhang [email protected]:
Could you please tell me where the instructions are?
— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/saiprabhakar/DeepDriving/issues/2#issuecomment-333279269, or mute the thread https://github.com/notifications/unsubscribe-auth/ALA9x1oPd23GJiKDZJLZeyaosNwRJQe-ks5snbMbgaJpZM4KjQJe .
I have install the development environment according to the instructions.Then I want to test the model.However,it seems the program doesn't work well.Maybe I omit some important steps.
My test steps is: 1.the default parameters of the traffic cars, the host car and the Caffe code is "three-lane" 2.the test data in the directory "Current_State_1F" is the default data from the original path. 3.I just run "sh ./torcs_test.sh".As a result,there are three windows coming into being.There are:"Image from leveldb","Error bar","Sematic Visualization".But just the "Image from leveldb" show an track image,other windows show whitespace. I also try to press P in a visulize window to start caffe, and then press C.It doesn't work. I don't know whether I describe the question clearly.Looking forward to you answer.Thank you very much.
I install the development environment like you, and everything is ok. However, the CNN drive the host car (chenyi) is unnormal, the car always left the lane and stop driving. Do you have any suggestion, @larenzhang .
Change the view of car can solve my question.
@larenzhang I am not sure what the problem is, since I dont work on this now. But I will look into it.
@larenzhang Hello, I have the same problem with you. Did you solve it?