wayveai / mile

PyTorch code for the paper "Model-Based Imitation Learning for Urban Driving".
MIT License
358 stars 34 forks source link

To run data collect script there is no anything. #18

Closed Tony-Hou closed 9 months ago

Tony-Hou commented 1 year ago

To run below script but specify carla_dataset directory no anything.

bash run/data_collect.sh /home/linjiehou/workspace/personal/auto/tools/carla/CarlaUE4.sh ./carla_dataset/ 2000 lb_data.yaml

[2023-06-08 21:42:16,608][utils.server_utils][INFO] - Killed Carla Servers on port 2000! [2023-06-08 21:42:17,629][utils.server_utils][INFO] - Killed Carla Servers on port 2000! [2023-06-08 21:42:18,635][utils.server_utils][INFO] - bash /home/linjiehou/workspace/personal/auto/tools/carla/CarlaUE4.sh -fps=25 -prefernvidia -quality-level=Epic -carla-rpc-port=2000 4.24.3-0+++UE4+Release-4.24 518 0 Disabling core dumps. MESA-INTEL: warning: Performance support disabled, consider sysctl dev.i915.perf_stream_paranoid=0

[2023-06-08 21:42:24,909][mile.agents.rl_birdview.rl_birdview_agent][INFO] - Resume checkpoint latest ckpt/ckpt_11833344.pth [2023-06-08 21:42:34,867][mile.agents.rl_birdview.rl_birdview_agent][INFO] - Loading wandb checkpoint: ckpt/ckpt_11833344.pth [2023-06-08 21:42:36,073][main][INFO] - Start from env_idx: 0, task_idx 0 wandb: Currently logged in as: hljzzu. Use wandb login --relogin to force relogin wandb: wandb version 0.15.4 is available! To upgrade, please run: wandb: $ pip install wandb --upgrade wandb: Tracking run with wandb version 0.13.9 wandb: Run data is saved locally in /home/linjiehou/workspace/personal/auto/mile/outputs/2023-06-08/21-42-16/wandb/run-20230608_214237-gdkxlru5 wandb: Run wandb offline to turn off syncing. wandb: Resuming run carla_dataset wandb: ⭐️ View project at https://wandb.ai/hljzzu/mile wandb: 🚀 View run at https://wandb.ai/hljzzu/mile/runs/gdkxlru5 /home/linjiehou/anaconda3/envs/carla/lib/python3.7/site-packages/gym/logger.py:30: UserWarning: WARN: Box bound precision lowered by casting to float32 warnings.warn(colorize('%s: %s'%('WARN', msg % args), 'yellow')) run/data_collect.sh: line 14: 255460 Killed python -u data_collect.py --config-name data_collect carla_sh_path=${CARLA_PATH} dataset_root=${DATASET_ROOT} port=${PORT} test_suites=${TEST_SUITE} PYTHON_RETURN=137!!! Start Over!!! [2023-06-08 21:48:38,139][utils.server_utils][INFO] - Killed Carla Servers on port 2000! 2000/tcp: 255512 /home/linjiehou/workspace/personal/auto/tools/carla/CarlaUE4.sh: line 5: 255512 Killed "$UE4_PROJECT_ROOT/CarlaUE4/Binaries/Linux/CarlaUE4-Linux-Shipping" CarlaUE4 $@ [2023-06-08 21:48:39,274][utils.server_utils][INFO] - Killed Carla Servers on port 2000! [2023-06-08 21:48:40,280][utils.server_utils][INFO] - bash /home/linjiehou/workspace/personal/auto/tools/carla/CarlaUE4.sh -fps=25 -prefernvidia -quality-level=Epic -carla-rpc-port=2000 4.24.3-0+++UE4+Release-4.24 518 0 Disabling core dumps. MESA-INTEL: warning: Performance support disabled, consider sysctl dev.i915.perf_stream_paranoid=0

beyondpzk commented 1 year ago

I have also met this problem, have you solved it? Thanks

anthonyhu commented 1 year ago

It takes a bit of time for the first run to be completed, I think 20mins+ then it will save the outputs to disk :)

airobotis commented 9 months ago

@anthonyhu

Thank you for this great work. I have some questions for collecting data process.

  1. Is it normal for CARLA to continuously be rebooted ?
  2. Even if I run collect bash script for more than 8 hours, the collected data output file is not saved. (CARLA 0.9.11) Is there anything I need to check?

I really appreciate if you can help me

anthonyhu commented 9 months ago

That's curious. The first run should be saved after 20mins. What GPU hardware are you using?

CARLA rebooting can sometimes happen, but it shouldn't be frequent.

airobotis commented 9 months ago

@anthonyhu Thank you for the information. I use NVIDIA GeForce RTX 2080 Super with Max-Q Design for laptop and NVIDIA GRID A100D-16C for desktop when I changed CARLA FPS from 25Hz to 10Hz, It works now.

anthonyhu commented 9 months ago

Great to hear! :)