ROS Human Detection

A human detection sensor setup for Social Aware Navigation.

To use GPU with docker while on AMD64 systems, install nvidia-container-toolkit with given instructions.

Docker based Installation

git clone https://github.com/AIResearchLab/ros-human-sensing.git

on AMD64

Startup Detection

cd src/ros-human-sensing/docker
docker compose -f compose.amd64.yml pull
docker compose -f compose.amd64.yml up

Stopping

docker compose -f compose.amd64.yml down

Remove docker volumes for resetting

docker volume rm docker_system

on JetsonNano

Startup

cd src/ros-human-sensing/docker
docker compose -f compose.jnano.yml pull
docker compose -f compose.jnano.yml up

Stopping

docker compose -f compose.jnano.yml down

Remove docker volumes for resetting

docker volume rm docker_system

ROS based Installation

1. Create a workspace

mkdir -p workspace/src
cd workspace/src

2. Clone packages

git clone https://github.com/KalanaRatnayake/yolo_ros.git
git clone https://github.com/KalanaRatnayake/detection_msgs.git
git clone https://github.com/KalanaRatnayake/boxmot_ros.git
git clone https://github.com/AIResearchLab/ros-human-sensing.git

3. Install dependencies

pip3 install -r yolo_ros/requirements.txt
pip3 install -r boxmot_ros/requirements.txt

4. Setup AIResearchLab/OrbbecSDK_ROS2, AIResearchLab/astra_legacy_ros or any other camera system that support standard ROS conventions following given instructions.

5. Build the packages

colcon build

6. Start the system

• Start the camera in one terminal

• Start Detection

ros2 launch ros-human-sensing detection.launch.py

• Start Detection and tracking

ros2 launch ros-human-sensing tracking.launch.py