Domain: Software Engineering / Robotics / Computer Vision (DIY Prototyping)
Persona: Senior Systems Integration Engineer & Lead Maker
2. Summarize (Strict Objectivity)
Abstract:
This technical project documentation details the end-to-end development of an autonomous catapult system. The project integrates mechanical engineering, embedded systems, and computer vision (CV). The hardware consists of a wooden frame with a high-tension bungee-cord power source and a custom metal hook-and-piston firing mechanism. The control system utilizes a Python-based CV model running on a host machine, which identifies specific objects (e.g., fruits or biological shapes) via a webcam feed. Upon detection, a signal is transmitted to a microcontroller that manages the firing cycle, including winch-driven arming and a software-timed release to ensure proper payload positioning. Testing involved kinematic calculations to estimate velocity and the fabrication of custom ballistic gel models to assess projectile impact and trajectory stability.
System Architecture and Development Summary:
00:00:25 Concept and Framing: The initial build focuses on a traditional wooden catapult frame. The structure uses basic lumber and a bungee-cord tension system for energy storage.
00:01:54 Computer Vision Integration: A Python script is implemented to process a webcam feed. The system leverages an existing object detection model (transfer learning) to identify specific classes of objects, such as dogs or fruits, and draw real-time bounding boxes for targeting.
00:02:48 Hardware-Software Bridge: The project transitions from a local PC script to an integrated system using a microcontroller housed in a custom electronics enclosure. This unit receives detection triggers from the CV script to initiate the firing sequence.
00:03:18 Firing Mechanism Iteration: The initial release design—a direct-pull piston—failed under high load. It was replaced by a mechanical hook-and-sear assembly inspired by firearm firing pins, which successfully handled the tension required for long-range launches.
00:04:33 Autonomous Logic and "Launch Zone": The firmware is programmed to monitor a specific "firing zone." The system remains in a primed state until the CV model confirms the presence of a pre-selected object within the detection coordinates.
00:05:41 Latency Management: A software timer was implemented within the release logic. This delay prevents "premature firing" by ensuring the payload has settled in the basket before the release mechanism is triggered.
00:07:25 Ballistic Testing & Prototyping: To test impact without risk to living subjects, 3D-printed molds were used to cast ballistic gel models. These models included internal salt-water bladders to simulate biological mass distribution and internal cavities.
00:08:19 Kinematic Analysis: Theoretical calculations were performed to determine work (measured in Joules) and expected exit velocity. Initial estimates suggested a velocity of approximately 19.8 m/s, though field tests showed significant variance due to mechanical friction and energy loss.
00:09:42 Roof-Top Deployment: The system was relocated to an elevated position (roof) to maximize potential energy and flight distance for the final validation tests.
00:11:13 Validation and Field Test: The final test confirmed successful autonomous detection and launch of the ballistic gel payload, validating the integration of the CV trigger with the mechanical release.