Control a camera platform (2DOF Revolute-Revolute Joints Robot) using camera feed to follow object. Using Hardware:

• Jetson nano
• Arduino
• 2x Servos
• CSI Camera Software:
• ROS2 humble
• Docker container

cd ~
git clone --recursive --depth=1 https://github.com/elsayedelsheikh/jetson-rrbotcam

This project is divided into 3 main parts:

• bringup_robot to communicate with the Arduino & control the hardware (2x Servos)
• object_detector to fire up the camera and detect the object to follow using various computer vision techniques (HSV, Deep learning)
• object_follower to control the hardware to follow the object centroid (x,y) using PID controller

• Uses ros_arduino_bridge to communicate with the Arduino through USB-Serial port
• Uses ros2_control to control the 2x Servos positions (Command & State interface) with custom Hardware interface

• Uses ros-deep-learning package to fire up the camera Multiple detectors are available:
• HSV detector node to detect a specific color range
• detectnet node to detect a specific object using deep learning models (SSD-Mobilenet by default) from jetson-inference

• Track a person while removing the background

./bringup_robot/bringup.sh

cd ~/jetson-inference
docker/run.sh --ros humble -v ~/jetson-rrbotcam/object_detector_follower:/object_detector_follower/
cd /object_detector_follower && colcon build --symlink-install
source /object_detector_follower/install/setup.bash

ros2 launch rrbot_cam_object_detector jetson_detector.launch
ros2 launch rrbot_cam_object_follower object_follower.launch.py