Project: "Person Follow"
Owner: "Anna Boschi"
Date: "2019/06"
The Jaffle is an upgrade of the Turtlebot3 waffle using as pc a Jetson Xavier Developer Kit and as camera an Intel Realsense d435i.
The goal of the project is to track and follow a person in an indoor enviroment using a small wheeled rover and suitable sensor to monitor the target person.
- Turtlebot3 waffle
- Jetson Xavier Developer Kit
- Realsense Camera d435i
- Battery LiPo 4S
- The launcher file jaffle_robot.launch launches the realsense d435i camera and both the core and the lidar of the jaffle.
- In the folder nodes there are the files python implemented for doing person tracking with Haarcascade classifier or Y.O.L.O. (both the code with the control implementation and the one with the obstacle avoidance basic implementation).
- In the folder src there are the Haarcascade classifiers used for the person detection.
|Version|Kinetic + Ubuntu Xenial| Melodic + Ubuntu Bionic|
- buildLibrealsense2Xavier
- realsense2_camera
- ddynamic_reconfigure
- jaffle_tracking
- hls_lfcd_lds_driver
- turtlebot3_msgs
- rosserial
- darknet_ros
- darknet_ros_person in order to have all the cfg and weight files of the retrained networks (Y.O.L.O. versions: 2, 3 and 3_Tiny retrained just for the class person) it is necessary to download also the repository darknet_ros_person of the PIC4SeRCentre and put the files in the related folders of the darknet_ros original repository.