--- markdown_generated_at: "2026-11-10" title: "Detect, track, score with Qualcomm Dragonwing Vision Kit" url: "https://www.qualcomm.com/developer/learning/first-robotics" --- # Detect, track, score with Qualcomm Dragonwing Vision Kit An open-source AI vision platform ready for FIRST® Robotics teams. ![](https://s7d1.scene7.com/is/image/dmqualcommprod/Frame?$QC_Responsive$&fmt=png-alpha) ![Rubik Pi 3 highlighting AI acceleration, low power consumption, UFS storage, and PhotonVision support.](https://s7d1.scene7.com/is/image/dmqualcommprod/rubik-x2?$QC_Responsive$&fmt=png-alpha) DEVELOPER KIT ## First Pi built on Qualcomm AI for vision. The RUBIK Pi 3 is a lightweight development board based on the Qualcomm® Dragonwing™ QCS6490. [Buy now](https://andymark.com/products/rubik-pi3-vision-bundle) [View documentation](https://first-rubik.github.io/) PHOTONVISION INSTALLATION ### Computer vision system for FIRST® Robotics competition PhotonVision comes ready to use on RUBIK Pi 3. Track targets, detect game pieces, and navigate autonomously. Just boot up and start configuring. [View installation guide](https://docs.photonvision.org/en/latest/docs/quick-start/quick-install.html#rubik-pi-3-installation) USING PHOTONVISION ### Run AI vision pipelines RUBIK Pi 3 uses Qualcomm's AI acceleration technology to deliver high-performance AprilTag processing and object detection. [View documentation](https://docs.photonvision.org/en/latest/docs/description.html) EDGE IMPULSE ### Train custom object detection models for your robots Edge Impulse is a Qualcomm platform that helps teams create, test, and train custom vision models to recognize specific game elements. Teams can build detection models and deploy them directly to the RUBIK Pi 3. [Login with FIRST ID](https://firstrobotics.edgeimpulse.com/login) [Edge Impulse features](https://first-rubik.github.io/docs/Edge-Impulse/features/) ABOUT ## The AI-Powered Vision System The FIRST® Robotics Competition (FRC) is an annual international high school robotics program that provides teams the opportunity to design, build, and compete with robotics and computer vision projects using the Qualcomm Dragonwing Vision Kit (QDVK). ![](https://s7d1.scene7.com/is/image/dmqualcommprod/AI-Powered?fmt=webp-alpha&qlt=85) ![](https://s7d1.scene7.com/is/image/dmqualcommprod/rubik-pi-logo?$QC_Responsive$&fmt=png-alpha) #### RUBIK Pi 3 QDVK comprises open-source tools and the RUBIK Pi 3, powered by the Qualcomm Dragonwing™ QCS640 processor. RUBIK Pi 3 features a fast neural processing unit (NPU) for AI-powered object detention directly on the device. It offers low power consumption, supports multiple cameras, and onboard UFS storage to support current and future vision use cases. [Learn more](https://www.thundercomm.com/rubik-pi-3/en/docs/about-rubikpi/) ![](https://s7d1.scene7.com/is/image/dmqualcommprod/edge-impulse-logo-1?$QC_Responsive$&fmt=png-alpha) #### Edge Impulse integration For creating and testing custom vision models, QDVK supports Edge Impulse, a Qualcomm platform for training models to recognize specific game elements. Teams can build detection models and deploy them directly to the RUBIK Pi 3. [Learn more](https://www.edgeimpulse.com/qualcomm) ![](https://s7d1.scene7.com/is/image/dmqualcommprod/photonvision-logo?$QC_Responsive$&fmt=png-alpha) #### PhotonVision support The QDVK also integrates PhotonVision, an easy-to-use open-source vision tool commonly used in the FRC community to track game pieces, field markers (e.g., AprilTags), and scoring targets. Using PhotonVision, teams can set up cameras to experiment with different vision configurations and approaches and see how vision data is processed on the QDVK during competition. [Learn more](https://docs.photonvision.org/en/latest/docs/description.html) ## Getting started ![](https://s7d1.scene7.com/is/image/dmqualcommprod/Layer_1?$QC_Glymph$&fmt=png-alpha) ### Registration Register for the FIRST® Robotics Competition and purchase the QDVK hardware module. [Register](https://www.firstinspires.org/programs/youth-registration) [Purchase](https://andymark.com/products/rubik-pi3-vision-bundle) ![](https://s7d1.scene7.com/is/image/dmqualcommprod/Layer_1-1?$QC_Glymph$&fmt=png-alpha) ### Set up Set up the vision system by following the camera setup tutorial, which walks through connecting cameras to the RUBIK Pi 3. [Learn more](https://first-rubik.github.io/docs/Intro/) ![](https://s7d1.scene7.com/is/image/dmqualcommprod/Layer_1-2?$QC_Glymph$&fmt=png-alpha) ### Build Download PhotonVision and start building basic vision functionality. [Installation guide](https://docs.photonvision.org/en/latest/docs/quick-start/quick-install.html) ![](https://s7d1.scene7.com/is/image/dmqualcommprod/Layer_1-3?$QC_Glymph$&fmt=png-alpha) ### Learn Explore Edge Impulse to create and test custom vision models. [Edge Impulse University Program](https://www.edgeimpulse.com/university) ## Benefits ![](https://s7d1.scene7.com/is/image/dmqualcommprod/switch-camera-1?$QC_Responsive$&fmt=png-alpha) #### AprilTag localization AprilTags are visual markers placed on the field that robots can easily detect with one or more cameras. They help robots understand where they are and line up accurately during autonomous and driver-controlled periods in FIRST Robotics competitions. ![](https://s7d1.scene7.com/is/image/dmqualcommprod/focus?$QC_Responsive$&fmt=png-alpha) #### Object detection on device Using Qualcomm's AI acceleration technology to deliver high-performance vision processing, QDVK ensures your robot can detect and analyze objects with precision and speed in both competition matches and the real world. ![](https://s7d1.scene7.com/is/image/dmqualcommprod/settings?$QC_Responsive$&fmt=png-alpha) #### Easy set up and integration The QDVK vision system works with PhotonVision and supports Edge Impulse for training and testing vision models. These familiar, open-source tools don't require custom hardware or a complicated setup, so there is more time to experiment with advanced computer vision capabilities. ![](https://s7d1.scene7.com/is/image/dmqualcommprod/layout-dashboard?$QC_Responsive$&fmt=png-alpha) #### Easy-to-use interface for vision basics PhotonVision helps teams handle common vision tasks like camera setup, tracking targets, and working with multiple cameras thanks to its easy-to-use interface. ## Resources ![](https://s7d1.scene7.com/is/image/dmqualcommprod/rubik-pi-3-unboxing-thumbnail?fmt=webp-alpha&qlt=85&wid=1800&fit=fit,1) #### Rubik Pi 3 intro & unboxing [Watch video](https://www.youtube.com/watch?v=1LRJVeAd1aU) ![](https://s7d1.scene7.com/is/image/dmqualcommprod/qualcomm-launcher-thumbnail?fmt=webp-alpha&qlt=85&wid=1800&fit=fit,1) #### Flash PhotonVision using Qualcomm Launcher Qualcomm® Launcher is a user‑friendly GUI tool that simplifies downloading and flashing operating systems onto Qualcomm® development kits. [See the documentation](https://first-rubik.github.io/docs/software_device/) ![](https://s7d1.scene7.com/is/image/dmqualcommprod/rubik-pi-3-object-detection-thumbnail?fmt=webp-alpha&qlt=85&wid=1800&fit=fit,1) #### Rubik Pi 3 Object Detection PhotonVision runs object detection on the Rubik Pi 3 by use of TensorflowLite, and a JNI code. [See the documentation](https://docs.photonvision.org/en/latest/docs/objectDetection/about-object-detection.html) ![](https://s7d1.scene7.com/is/image/dmqualcommprod/rubik-pi-3-first-robotics-thumbnail?fmt=webp-alpha&qlt=85&wid=1800&fit=fit,1) #### Build Along Rubik Pi 3 and First Robotics Learn how to use PhotonVision, an AprilTag/object tracking framework for your RUBIK PI project. [View in YouTube](https://www.youtube.com/watch?v=QSTxjy1WHPQ) ## Discover our latest insights Developer Blog [Detect Jenga blocks in real-time: a hands-on guide with Edge Impulse and TensorFlow Lite](https://www.qualcomm.com/developer/blog/2025/09/detect-jenga-blocks-real-time-hands-on-guide-edge-impulse-tensorflow-lite) Sep 22, 2025 [Edge AI](https://www.qualcomm.com/developer/blog?tags=Edge%20AI) [Windows on Snapdragon](https://www.qualcomm.com/developer/blog?tags=Windows%20on%20Snapdragon) Developer Blog [Meet Arduino UNO Q: Qualcomm Dragonwing fueled AI in a blink](https://www.qualcomm.com/developer/blog/2025/10/meet-arduino-uno-q-qualcomm-dragonwing-fueled-ai-in-a-blink) Oct 7, 2025 [Internet of Things](https://www.qualcomm.com/developer/blog?tags=Internet%20of%20Things) [Partner](https://www.qualcomm.com/developer/blog?tags=Partner) [Developer](https://www.qualcomm.com/developer/blog?tags=Developer) ## Connect with our communities [LinkedIn](https://www.linkedin.com/showcase/qualcomm-developer) [See what's new](https://x.com/Qualcomm_Dev) [YouTube](https://www.youtube.com/qualcommdev) [GitHub](https://github.com/qualcomm) ![](https://s7d1.scene7.com/is/image/dmqualcommprod/discord-icon-1?$QC_Responsive$&fmt=png-alpha) [Discord](https://discord.com/invite/qualcommdevelopernetwork) [Visit the forum](https://mysupport.qualcomm.com/supportforums/s/)