BraccioPlayMemory

Getting started

• Install http://wiki.ros.org/kinetic
• Flash an Arduino Uno with the Arduino code
• Run

$ catkin_make

in the /catkin_ws folder

• If everything is installed correctly, open 5 terminals
• Source the workspace in every terminal

source devel/setup.bash

• Terminal 1:

$ roscore

• Terminal 2: Use usb_cam to publish webcam images, darknet_ros would subscribe it to

$ roslaunch usb_cam usb_cam-test.launch

• Terminal 3: Start YOLOv2 object detection

$ roslaunch darknet_ros Braccio_darknet_ros.launch

• Terminal 4: Transfer output from YOLOv2 into robot instruction

$ rosrun detect_and_collect parse_and_publish

• Terminal 5: Connect to Arduino via serial connection

$ rosrun rosserial_python serial_node.py /dev/ttyACM0

Our work:

• Configuration of the network in /darknet/cfg/yolo-obj.cfg
• Dataset and labels in /darknet/data/obj
• ROS code in /catkin_ws/src/detect_and_collect
• Code on the Arduino in /Arduino/braccio_ros/braccio_ros.ino
• Final weights in /darknet/TrainingTinyTiny/yolo-obj-1500.weights

Work of others

• Labeling Tool
• Visualization Tool
• YOLO
• Darknet