# Dog Catapult Source: https://www.youtube.com/watch?v=oA85M9JHsW0 Recap page: https://rapidrecap.app/video/oA85M9JHsW0 Generated: 2026-02-13T17:37:38.953+00:00 --- ## Quick Overview The creator successfully built and tested an automated catapult, nicknamed the "Dog Chucker 9000," designed to launch objects based on object detection (specifically targeting dogs, bananas, and apples), ultimately testing it by launching a ballistic gel dog mold filled with fake blood, which resulted in a messy but successful demonstration of the system's functionality. **Key Points:** - The project involved building a large, wooden, wheeled catapult (the "Dog Chucker 9000") powered by bungee cords. - The system was automated using computer vision (YOLOv8s model) programmed in Python to detect specific objects like 'dog', 'banana', and 'apple' in the launch zone to trigger the firing mechanism. - The creator consulted with Mark Rober via video call for physics calculations, which estimated a potential launch distance of 19.8 meters (about 65 feet). - To test the system safely indoors, the creator poured a ballistic gel mixture, including a small cat skull, into a 3D-printed dog mold to simulate the dog projectile. - The ballistic gel dog was successfully launched indoors and exploded upon hitting a backstop, demonstrating the detection and launch mechanism worked. - The final outdoor test involved launching a watermelon, which hit a target board on the garage door, causing significant splatter. - The creator ultimately decided against launching the ballistic gel dog outdoors due to safety concerns, especially given the controversy surrounding a previous YouTuber incident involving an opossum. ![Screenshot at 00:56: The creator presents the finished wooden catapult structure in the garage, holding a hammer and ready to explain the device's mechanism.](https://ss.rapidrecap.app/screens/oA85M9JHsW0/00-00-56.jpg) **Context:** This video documents the process of building and testing a large, motorized, computer-vision-controlled catapult, humorously dubbed the "Dog Chucker 9000." The creator initially sought input from engineering YouTuber Mark Rober regarding the physics calculations required to estimate the projectile's trajectory. The core innovation is the integration of object detection software to automatically arm and fire the catapult only when a specified target object (initially dogs, later changed to fruit and the gel dog) is present in the firing cup. ## Detailed Analysis The video details the construction and testing of a custom-built, wheeled catapult designed to launch objects based on object detection. The creator first researches how to build a catapult, seeking advice from Mark Rober on physics calculations, which estimated a launch distance of 19.8 meters for a 6-pound projectile. The catapult frame is constructed primarily from wood and utilizes yellow bungee cords for power. The firing mechanism is automated via a Python script using OpenCV and a YOLOv8s model to detect objects. The creator defines target objects in the code, initially including 'dog', 'banana', and 'apple', and later changing the target to 'dog' after consulting with Mark Rober about the ethics of launching a dog. To test the system safely indoors, the creator constructs a ballistic gel replica of a dog inside a 3D-printed mold, filling it with gel and a small skull, and then tests the device by launching it, which works successfully, causing the gel dog to splatter. After the successful indoor test, the creator moves the catapult outdoors for a higher-power test, launching a watermelon which hits a target board, creating a large red splatter. The creator ultimately decides against testing the ballistic gel dog outdoors due to safety concerns and the controversy surrounding a similar incident involving another YouTuber catapulting an opossum. ### Project Concept & Physics Consultation - The goal is a catapult that launches objects automatically based on object detection - Consulted Mark Rober for physics calculations, yielding an estimated launch distance of 19.8m for a 6lb projectile using 1800J of energy. ### Catapult Construction & Automation - Built a large wooden catapult frame on wheels - Incorporated a mechanical trigger mechanism released by cutting a string, controlled by custom electronics (ESP32/Arduino setup) and a solenoid valve. ### Object Detection Implementation - Wrote Python code using OpenCV and YOLOv8s to detect objects in real-time - Defined targets in code like 'dog', 'banana', 'apple', and later 'doggie' and 'watermelon' to trigger the launch. ### Indoor Testing & Gel Mold - Built a ballistic gel dog mold (3D printed, coated with epoxy) filled with gel and a small skull to simulate the projectile - Successfully launched the gel dog indoors, causing it to splatter. ### Outdoor Test & Final Decision - Moved the catapult outdoors for a full-scale test, successfully launching a watermelon which hit a target board - Decided against launching the dog mold outdoors due to safety and ethical concerns raised by the controversy surrounding another YouTuber's opossum launch. ![Screenshot at 00:18: The creator driving, explaining that the next step involves buying wood to build the catapult.](https://ss.rapidrecap.app/screens/oA85M9JHsW0/00-00-18.jpg) ![Screenshot at 00:35: The creator researching historical catapult designs on Google Images while in Home Depot.](https://ss.rapidrecap.app/screens/oA85M9JHsW0/00-00-35.jpg) ![Screenshot at 00:56: The creator standing proudly next to the completed, large wooden catapult structure in the garage.](https://ss.rapidrecap.app/screens/oA85M9JHsW0/00-00-56.jpg) ![Screenshot at 02:09: The Python script running, showing the webcam feed with an object detection window displaying the creator detected as 'person'.](https://ss.rapidrecap.app/screens/oA85M9JHsW0/00-02-09.jpg) ![Screenshot at 03:33: A close-up of the custom metal trigger mechanism inspired by firearm firing pins.](https://ss.rapidrecap.app/screens/oA85M9JHsW0/00-03-33.jpg)