edge-conde's Introduction

local_mapping_ros

Local mapper package. The role of the local mapper is to take the RGB-D camera images and output the cone classes and positions in 3D space.

Configurations such as tf names or topics can be updated in the launch file of this package.

Inputs

/camera/color/image_raw (sensor_msgs/Image): Color image
/camera/depth/image_rect_raw (sensor_msgs/Image): Depth image
/camera/depth/camera_info (sensor_msgs/CameraInfo): Depth camera intrinsic parameters

Outputs

/cones (sensor_msgs/ConeArray): Cone array in car axis
/local_cone_markers (visualization_msgs/MarkerArray): Array of markers for visualization in RViz or Foxglove

Requirements

ROS 2 Humble
CMake
LibTorch
lart_msgs
C++17 compiler

Docker

Building

If you aren't working in this package and just want to use it, or deploying to production, Docker is recommended (takes a lot of time the first time but shouldn't break).

Bare metal

Install the dependencies listed above.
Make a new directory for your workspace.
- Example: mkdir -p ~/catkin_ws/src
Put this directory in the src directory of the workspace.
Run the command source /opt/ros/humble/setup.bash.
From the root of the workspace run:
- rosdep install -i --from-path src --rosdistro humble -y
- colcon build

Docker

Run the command docker build -t local_mapping . and wait about some centuries.
Run the command docker run --gpus all local_mapping.

edge-conde's People

Contributors

Stargazers

Watchers

edge-conde's Issues

Validate the depth range

Validate the depth range of the local mapper (including the camera's depth sensing and the cone detection range).
Consider cones are 5 meters apart and the car needs to "see" at least 2 cones in front on each side.

Write local mapper documentation

Write rigorous local mapper documentation, including relevant equations and explicit and detailed diagrams.

Gather the TorchScript model from the package

The TorchScript model is supposed to be gathered from the package's "model" directory. Instead, it is being gathered from the workspace.
Also, write documentation explaining where to put the model and how to configure the name.

Recommend Projects