adamzdw / CICT

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Continuous Intention and Continuous Trajectory

arXiv License: MIT

robocar

Sensors used:

Learning Module

  • Learning models
    • Driving Intention Module: learning/models.py GeneratorUNet and Discriminator
    • Trajectory Generation Module: learning/path_model.py ModelGRU
  • Train model
python scripts/carla/train_path.py
python scripts/carla/train_GAN.py

Install

cd scripts
bash install.sh

Features

  • Camera image reading in python with pybind11
bash device/camera/build.sh
python device/camera/run.py
  • LiDAR data reading
cd device/lidar/velodyne-driver
mkdir build
cd build
cmake ..
make -j16
sudo make install
cd ../../
python lidar.py
  • IMU data reading
bash scripts/real/get_permission.sh
python device/imu/mtnode.py
  • XBox control
    • Right axis up: speed
    • Top axis left and right: rotation
    • Buttom Y: forward
    • Buttom A: backward
    • Buttom LOG: break and stop
    • Hat up and down: increase or reduce max speed
    • Hat right and left: increase or reduce acceleration
bash scripts/real/get_permission.sh
cd device/controller
python xbox.py

ROS Wrapper

Build

cd ROS
bash build.sh

Launch

source devel/setup.bash
roslaunch mynteye_wrapper_d mynteye.launch
roslaunch velodyne_pointcloud VLP16_points.launch
python src/gps/scripts/run_gps.py
python src/imu/scripts/mtnode.py

Record

rosbag record /mynteye/left/image_color /velodyne_points /gps /imu

Read Rosbag

python ROS/src/tools/save_img.py -d 1 -b 2020-07-11-17-50-49.bag

Calibration

cd ROS
bash run_collect.sh

Robot Parameter

  • Shaft length : 688.008mm
  • Axle distance :1448.555mm
  • Voltage : 48V

Video

video

Citation

If you use our source code, please consider citing the following:

@article{wang2020imitation,
  title={Imitation Learning of Hierarchical Driving Model: from Continuous Intention to Continuous Trajectory},
  author={Wang, Yunkai and Zhang, Dongkun and Wang, Jingke and Chen, Zexi and Wang, Yue and Xiong, Rong},
  journal={arXiv preprint arXiv:2010.10393},
  year={2020}
}

About

License:MIT License


Languages

Language:Python 63.1%Language:C++ 29.2%Language:CMake 3.9%Language:MATLAB 3.7%Language:Shell 0.2%