AI Video Summary: Automatic Bullseye, MOVING Dartboard
Channel: Mark Rober
TL;DR
Mark Rober demonstrates an automated moving dartboard engineered over three years to guarantee bullseyes or intentionally miss. The system uses a VICON motion capture setup to predict dart trajectories via parabolic physics and moves the board with high-precision stepper motors in under 200 milliseconds.
Key Points
- — Mark Rober introduces the project, noting it took over three years of engineering work with a former NASA colleague.
- — The system relies on two main tasks: predicting the dart's landing spot within 200ms and moving the board to that location in another 200ms.
- — Prediction is achieved using a VICON motion capture system with 6 cameras tracking retroreflectors on the dart to determine 3D position.
- — The software uses MATLAB to calculate the dart's path, treating vertical motion as a parabola and horizontal motion as a linear equation.
- — The board moves using linear sliders and six stepper motors connected by fishing line to achieve sub-millimeter precision.
- — The system refines its prediction 10 to 100 times in less than half a second to ensure the board jitters into the exact final position.
- — The team tested the board in a bar, where drunk patrons were thrilled to suddenly become expert dart players.
Detailed Summary
Mark Rober presents an automated moving dartboard that he and a former NASA colleague spent over three years developing. The device is designed to either guarantee a bullseye for the user or intentionally move away to ensure a miss. The core challenge of the project involves two distinct phases: predicting where a thrown dart will land and physically moving the board to that spot, both of which must occur within a total window of approximately 400 milliseconds. To predict the dart's trajectory, the team utilizes a VICON motion capture system consisting of six high-speed cameras. The dart is equipped with tiny retroreflectors that bounce infrared light back to the cameras, allowing the system to track the object's XYZ position in 3D space at 260 frames per second. Using MATLAB, the software analyzes this data to predict the landing spot. The vertical motion is calculated based on the physics of a perfect parabola, while the horizontal motion is treated as a linear path using the equation y = mx + b. Once the target location is determined, the board is moved using a system of linear sliders and six stepper motors. These motors, controlled by an Arduino-compatible board and powered by preamps, wind and unwind fishing lines attached to the center of the board to translate it precisely. The system updates its prediction 10 to 100 times during the throw, allowing for sub-millimeter accuracy. The video concludes with footage of the board being tested in a bar, where intoxicated patrons were amazed to find themselves hitting bullseyes consistently.
Tags: engineering, dartboard, motion capture, robotics, physics, vicon, stepper motors, mark rober