Skip to content

Latest commit

 

History

History
98 lines (63 loc) · 1.63 KB

ros_related.md

File metadata and controls

98 lines (63 loc) · 1.63 KB
Raw

Extra ROS Related Tools

Assuming that you have already installed ROS using the ROS-desktop variant, the following can help you install all the required software packages.

CUDA SGM

  1. Install PCL_ROS and Image_View
sudo apt-get update
sudo apt install ros-melodic-pcl-ros ros-melodic-image-view
  1. Build CUDA ROS SGM Package
cd /../cuda_sgm_ros
catkin bt

Apart from some deprecated syntax warnings, it should build fine.

  1. Download test data
a) Get the original bag file from this folder:
   https://drive.google.com/open?id=1Qplh9fYt10t5H1p_adJkjhk3T7CjE2Xw
b) Also download the calibration file
   file: fisheye_stereo_rectify.yml
  1. Move the calibration file to the .ros folder
mv fisheye_stereo_rectify.yml ~/.ros/
  1. Re-map image topics and launch stereo_gpu.launch
roslaunch cuda_sgm_ros stereo_gpu.launch
  1. Visualize
# Disparity View
rosrun image_view disparity_view image:=/stereo_gpu/disparity
# Disparity Image
rqt_image_view

Some benchmarks on the CUDA Semi Global Matching ROS Node: CUDA_Benchmark

Creating an isolated ROS Workspace for Python 3.5 Deep Learning

  1. Create a workspace
cd ~
mkdir ros_py35
cd ros_py35
  1. Unset existing ROS workspaces if you have any
vim ~/.bashrc
  1. Get Python3 helpers
sudo pip3 install rosdep
sudo pip3 install rosinstall_generator
sudo pip3 install wstool
sudo pip3 install rosinstall
  1. Configure your catkin workspace
mkdir src
catkin config --init
catkin config -DCMAKE_BUILD_TYPE=Release