LAV

原repo地址：https://github.com/dotchen/LAV

原paper: Learning from all vehicles

此处仅作为参考对比代码，更多讨论见discussion部分！Please click discussion section to discussion on this task.

Getting Started

• To run CARLA and train the models, make sure you are using a machine with at least a mid-end GPU.

• Please follow INSTALL.md to setup the environment.

  1. clone 与 git_lfs下载

     git clone --recurse-submodules git@github.com:Kin-Zhang/LAV.git
     curl -s https://packagecloud.io/install/repositories/github/git-lfs/script.deb.sh | sudo bash

  2. 环境 与 CUDA安装 11.3

     conda env create -f environment.yaml
     conda install pytorch torchvision torchaudio cudatoolkit=11.3 -c pytorch
     conda install pytorch-scatter -c pyg

• We also release our LAV dataset. Download the dataset HERE. [还没有开数据集是如何收集的，本来以为不太多 打算试试，然后一看 emmm 太多label了 还是等作者开把... ]

Training

We adopt a LBC-style staged privileged distillation framework. Please refer to TRAINING.md for more details.

以下为简单版复制阶段：

1. Privileged Motion Planning

   python -m lav.train_bev

2. Semantic Segmentation

   python -m lav.train_seg

3. RGB Braking Prediction

   python -m lav.train_bra

4. Point Painting, 主要是针对前面训出来的 对数据集进行添加

   python -m lav.data_paint

5. Perception Pre-training

   python -m lav.train_full --perceive-only

6. End-to-end Training

   python -m lav.train_full

Evaluation

We additionally provide examplery trained weights in the weights folder if you would like to directly evaluate. They are trained on Town01, 03, 04, 06. Make sure you are launching CARLA with the -vulkan flag.

运行 ./leaderboard/scripts/run_evaluation.sh 其中文件可改为此处

#!/bin/bash

#!改这两个地址=====
export CARLA_ROOT=/home/kin/CARLA
export LAV=/home/kin/lav

export LEADERBOARD_ROOT=${LAV}/leaderboard
export SCENARIO_RUNNER_ROOT=${LAV}/scenario_runner
export PYTHONPATH=$PYTHONPATH:"${CARLA_ROOT}/PythonAPI/carla/":"${SCENARIO_RUNNER_ROOT}":"${LEADERBOARD_ROOT}":${CARLA_ROOT}/CARLA/PythonAPI/carla/dist/carla-0.9.11-py3.7-linux-x86_64.egg
export TEAM_AGENT=${LAV}/team_code/lav_agent.py
export TEAM_CONFIG=${LAV}/team_code/config.yaml

export SCENARIOS=${LEADERBOARD_ROOT}/data/all_towns_traffic_scenarios_public.json
export ROUTES=${LEADERBOARD_ROOT}/data/routes_devtest.xml
export REPETITIONS=1
export CHECKPOINT_ENDPOINT=results.json
export DEBUG_CHALLENGE=0
export CHALLENGE_TRACK_CODENAME=SENSORS

python3 ${LEADERBOARD_ROOT}/leaderboard/leaderboard_evaluator.py \
--scenarios=${SCENARIOS}  \
--routes=${ROUTES} \
--repetitions=${REPETITIONS} \
--track=${CHALLENGE_TRACK_CODENAME} \
--checkpoint=${CHECKPOINT_ENDPOINT} \
--agent=${TEAM_AGENT} \
--agent-config=${TEAM_CONFIG} \
--debug=${DEBUG_CHALLENGE} \
--record=${RECORD_PATH} \
--resume=${RESUME}

Use ROUTES=assets/routes_lav_valid.xml to run our ablation routes, or ROUTES=leaderboard/data/routes_valid.xml for the validation routes provided by leaderboard.

Acknowledgements

We thank Tianwei Yin for the pillar generation code. The ERFNet codes are taken from the official ERFNet repo.

License

This repo is released under the Apache 2.0 License (please refer to the LICENSE file for details).