This reposity is built with Catkin on Ubuntu 18.04 and has additional dependencies on CUDA10, TensorRT, and OpenCV 3.4. For best results, it is recommended to use NVIDIA Driver 410 to build CUDA with TensorRT version 5.0.2.6, as these are the versions used on the MIT/DUT18D car.

CUDA is an NVIDIA GPU programming language, with installation instructions that can be found on NVIDIA's website at: https://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html

TensorRT is a CUDA-based deep learning inference platform optimized for NVIDIA GPU's, used to run cone detections and keypoint detections on the MIT/DUT18D car. TensorRT installation instructions can also be found on NVIDIA's website at: https://docs.nvidia.com/deeplearning/sdk/tensorrt-install-guide/index.html

OpenCV is an open-source computer vision library, used in this repository for processing images and storing basic image-related data structures. OpenCV 3.4 can be installed by running the following commands:

git clone git@github.com:opencv/opencv_contrib.git sudo apt-get install -y qtbase5-dev qtdeclarative5-dev cd opencv_contrib git checkout 3.4 cd .. git clone https://github.com/opencv/opencv.git cd opencv git checkout 3.4 mkdir build cd build cmake -DCMAKE_BUILD_TYPE=RELEASE -DCMAKE_INSTALL_PREFIX=/usr/local -DWITH_CUDA=ON -DENABLE_FAST_MATH=1 -DCUDA_FAST_MATH=1 -DWITH_CUBLAS=1 -DBUILD_opencv_cudacodec=OFF -DWITH_QT=ON -DWITH_OPENGL=ON -DFORCE_VTK=OFF -DWITH_TBB=OFF -DWITH_MKL=ON -DMKL_USE_MULTITHREAD=ON -DMKL_WITH_TBB=ON -DWITH_IPP=ON -DWITH_GDAL=ON -DWITH_XINE=OFF -DBUILD_EXAMPLES=OFF -DCUDA_ARCH_PTX="" -DCUDA_ARCH_BIN="6.1" -DOPENCV_EXTRA_MODULES_PATH=../../opencv_contrib/modules .. make -j"$(nproc)" sudo make install