Skip to content

Latest commit

 

History

History
133 lines (100 loc) · 4.22 KB

quickstart.md

File metadata and controls

133 lines (100 loc) · 4.22 KB

Quickstart

  • Minimal environment for evaluation

    conda create -n crowdbot_eval python=3.8
conda activate crowdbot_eval
## or `source activate crowdbot_eval` in ubuntu
python -m pip install -U pip
sudo apt-get install ros-$ROS_DISTRO-ros-numpy
sudo apt-get install ros-$ROS_DISTRO-tf2-sensor-msgs

## install using requirements (recommended)
pip3 install -r requirements_eval.txt

## install using scripts (optional)
python -m pip install --user numpy scipy matplotlib ipython jupyter pandas sympy nose
python -m pip install --user rospkg pycryptodomex python-gnupg
# https://github.com/moble/quaternion
python -m pip install --upgrade --force-reinstall numba==0.53.1
python -m pip install --upgrade --force-reinstall numpy-quaternion

## install qolo
python -m pip install -e .

Simplified commands for evaluation

  1. Set direcroty in data/data_path.yaml

    bagbase_dir: /hdd/data_qolo/crowd_qolo_recordings/
outbase_dir: /hdd/data_qolo/crowdbot_data_analysis/

    • bagbase_dir is where rosbags are stored

    • outbase_dir is where extracted lidar, qolo_state, algorithm result, and evaluation results are stored, In terms of example, please check outbase_dir

  2. data conversion from original rosbags

    Before exporting lidar files, please check whether PyKDL is installed successfully. If not, please check pykdl_installation_guide.md

    • export pointcloud and lidar timestamp

      python3 qolo/gen_lidar_from_rosbags.py --overwrite -f 0410_shared_control
      • --overwrite flag is used to overwrite existing data
      • --compressed flag is used to save pointcloud as compressed pcd files (.pcd). If not, save with npy-like format
      • -f flag is used to specify data folder

    Tips: For following steps, you can easily execute a single shell script as:

    cd sh_scripts
sh data_export_eval_source_data.sh -e=py38cuda110 -t=0410_shared_control

    • export qolo status

      python3 qolo/commands2npy.py --overwrite -f 0410_shared_control
python3 qolo/tfqolo2npy.py --overwrite -f 0410_shared_control
python3 qolo/twist2npy.py --overwrite -f 0410_shared_control
python3 qolo/pose2d2npy.py --overwrite -f 0410_shared_control

  3. apply algorithms to extracted data

    python3 qolo/gen_detection_res.py -f 0410_shared_control
python3 qolo/gen_tracking_res.py -f 0410_shared_control

  4. visualization of current evaluation result

    python3 qolo/gen_viz_img.py -f 0410_shared_control
python3 qolo/gen_animation.py -f 0410_shared_control

    The visualization results can be found in ./data/0410_shared_control_processed/media/

  5. evaluate the qolo and crowd data from algorithm result and extracted data

    • default: skip the sequences that have been analyzed

    • overwrite all produced data

      python3 qolo/eval_qolo_path.py --overwrite -f 0410_shared_control
python3 qolo/eval_qolo_ctrl.py --overwrite -f 0410_shared_control
python3 qolo/eval_crowd.py --overwrite -f 0410_shared_control
      • --overwrite flag is used to overwrite existing data
      • --replot flag is used to update plots for each evaluation pipeline

    The visualization results can be found in ./data/0410_shared_control_processed/metrics/

outbase_dir example

taking 0424_mds_processed/ as an example

$ tree -L 2
.
├── alg_res
│   ├── detections
│   └── tracks
├── lidars
│   ├── 2021-04-24-12-04-04
│   ├── 2021-04-24-12-07-57
│   ├── 2021-04-24-12-10-45
│   ├── 2021-04-24-12-54-04
│   ├── 2021-04-24-12-56-59
│   └── 2021-04-24-13-03-39
├── metrics
│   ├── 2021-04-24-12-04-04
│   ├── 2021-04-24-12-07-57
│   ├── 2021-04-24-12-10-45
│   ├── 2021-04-24-12-54-04
│   ├── 2021-04-24-12-56-59
│   └── 2021-04-24-13-03-39
└── source_data
    ├── commands
    ├── pose2d
    ├── tf_qolo
    ├── timestamp
    └── twist