Bin Boy

An autonomous trash can

Install

sudo apt update && sudo apt install ros-foxy-cv-bridge ros-foxy-vision-msgs -y
sudo apt install libopencv-dev python3-opencv

Camera (csi_camera_cpp)

GStreamer -> /raw_image (unique_ptr)
ROS msg --cv_bridge--> OpenCV compatible viewing
Minimize latency with unique_ptr (no copying w/n nodes in same process)
Person detector subscribes to /image_raw running pre-trained MobileNet-SSD Caffe model performing inference

source install/setup.bash

# defaults: detector ON, skip=1, annotated image OFF

ros2 launch csi_camera_cpp csi_camera_ipc.launch.py

ros2 launch csi_camera_cpp csi_camera_ipc.launch.py run_detector:=false

ros2 launch csi_camera_cpp csi_camera_ipc.launch.py detection_frame_skip:=4 publish_annotated_image:=false

Debug

sudo systemctl <status|restart> nvargus-daemon

gst-launch-1.0 nvarguscamerasrc ! fakesink