Home

NURDVision

Jetson Setup | Jetson Install | PC Install | Notes and Info | FRC Robot Implementation

SuperNURDs implementation of vision co-processing

This code was primarily developed for use on a NVIDIA Jetson TK1 connected over a network to a FIRST Robotics Competition Robot.

Using the OpenCV Libraries, NURDVision takes the stream of a camera and processes the image to find two targets that match the parameters set in the config.h file. Bounding boxes are then drawn over the image enabling the program to track and process the targets' position, angle, and amount of offset from center.

NURDVision uses WPI Utility and the WPI NetworkTables Protocol to make found targets' position, angle, and offset viewable over a network in the form of a NetworkTable that can be read by any device using the NetworkTable protocol.

By using the WPI CameraServer Library, either the processed or raw camera image streams (only one is viewable at a time and can be toggled) can be viewed over the network by connecting to http://deviceIP:portNumber/stream.mjpg. The portNumber is defined in the config file and defaults to 1180

---

NOTE: This is the runnable program and setup of NURDVision, to see a simple implementation example on an FRC Robot visit: NURDVisionRobot GitHub

Major resource used - SMblyRequired