Virtual Lane Boundary Generation for Human-Compatible Autonomous Driving: A Tight Coupling between Perception and Planning
Our code is released under a GPLv3 license. In this code, we aim to recognize, generate and track virtual lane boundaries to help navigate autonomous vehicles in urban scenarios. It has been tested on Ubuntu 16.04LTS and should work on 16.04 or newer versions. We extend our previous work in
@INPROCEEDINGS{li2019virtual,
title={Virtual Lane Boundary Generation for Human-Compatible Autonomous Driving: A Tight Coupling between Perception and Planning},
author={Li, Binbin and Song, Dezhen and Ramchandani, Ankit and Cheng, Hsin-Min and Wang, Di and Xu, Yiliang and Chen, Baifan},
booktitle={2019 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)},
pages={3733--3739},
year={2019},
organization={IEEE}
}
@inproceedings{li2018lane,
title={Lane marking quality assessment for autonomous driving},
author={Li, Binbin and Song, Dezhen and Li, Haifeng and Pike, Adam and Carlson, Paul},
booktitle={2018 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)},
pages={1--9},
year={2018},
organization={IEEE}
}
If you use our code in an academic work, please cite the aforementioned paper.
git clone --recurse-submodules [email protected]:bli-tamu/LDRT.git
You can use ./install-dependencies.sh for convenience, or do it manually as follows:
sudo apt-get install --no-install-recommends \
build-essential \
cmake \
git \
wget \
libatlas-base-dev `#For Caffe` \
libboost-all-dev \
libgflags-dev \
libgoogle-glog-dev \
libhdf5-serial-dev \
libleveldb-dev \
liblmdb-dev \
libopencv-dev \
libprotobuf-dev \
libsnappy-dev \
protobuf-compiler \
python3-pip \
python3-dev \
libmatio-dev `#MAT File I/O` \
libpugixml-dev `#XML processing` \
libalglib-dev `#Numerical analysis` \
libgsl-dev `#GNU scientific library` \
libopenblas-dev `#Linear algebra library` \
libeigen3-dev \
libpcap-dev `#Packet capture library`
pip3 install numpy matplotlib pandas scipy pybind11
You can use ./build.sh for default build, or do it manually as follows:
-
ICNet
Copy the example configuration file and apply patch.
cp 3rdParty/ICNet/PSPNet/Makefile.config.example 3rdParty/ICNet/PSPNet/Makefile.config patch -p0 -i 3rdParty/patches/ICNet.patchModified the configuration file before compile if needed.
cd 3rdParty/ICNet/PSPNet make -
This project uses CMake (http://www.cmake.org), a cross-platform build system.
mkdir build && cd build cmake .. make
- Install VINS-Fusion to generate visual odometry
vio.txtfrom the dataset, and put it underdemo/data/datasetNamefolder. - Download GPS way points from Google maps, utilize the script
matlab/process_KITTIdata.mto generate the map, and put the data in thedemo/data/datasetName/priorMap.txt.
- We use KITTI dataset to demostrate our code,
prepDemo.shis a script that downloads and prepares the demo. For example,./prepDemo.sh 2011_09_26_drive_0056_syncwill automatically prepare the PSPNet models, 2011_09_26_drive_0056_sync dataset and its calibration files. - Run the dataset with VINS-Fusion and place the
vio.txtindemo/data/datasetName.
- Run with
runDemo.sh, for example,./runDemo.sh 2011_09_26_drive_0056_syncwill run the code and keep a log file indemo/data/2011_09_26_drive_0056_sync/.
- Binbin Li [email protected]
- Ankit Ramchandani [email protected]
- Di Wang [email protected]
- Aaron Kingery [email protected]
- Aaron Angert [email protected]
- Dezhen Song [email protected]
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent