OptiTrack

Engineering the perfect putt Building a machine that outclasses human skill is rarely a matter of simple assembly. When Shane Wighton set out to build an Auto-Correcting Mini Golf Club, he entered a territory where physics and software latency collide. The objective was straightforward: create a putter that adjusts its face angle during a swing to ensure the ball always finds the hole. In practice, this required solving mechanical paradoxes that would make a standard PC build look like child's play. The project demands a total system response time of less than 200 milliseconds—roughly the duration of a human blink—to track, calculate, and execute a physical correction. Mechanical locks and torque cancellation A primary hurdle in precision hardware is ensuring the mechanism doesn't flex or yield under pressure. Wighton initially struggled with the "springiness" of a standard drive shaft. His solution involved a Worm Gearbox, a one-way mechanical advantage system where a spinning screw drives a larger gear. This setup ensures that when the ball strikes the putter head, the force cannot back-drive the motor, providing a rock-solid hitting surface. However, moving a heavy putter head at high speeds creates a massive reaction torque that would twist the club right out of a user's hands. While an initial attempt at an Inertial Counterweight failed due to lateral acceleration issues, a deeper dive into the math revealed a more elegant fix. By adding Inertia Rings to the ODrive controlled motor itself, Wighton utilized the motor’s own internal acceleration to perfectly cancel out the putter's twist. This kind of pragmatic optimization—using a secondary problem to solve a primary one—is the hallmark of high-level engineering. Precision tracking in three dimensions Software cannot correct what it cannot see. To feed data into his Python scripts, Wighton deployed a suite of twelve OptiTrack cameras. These sensors utilize infrared pulses to locate reflective markers with sub-millimeter accuracy. The challenge extended to the ball itself; standard golf balls don't reflect infrared light reliably. The fix was low-tech but effective: wrapping the ball in high-visibility reflective tape to turn it into a singular, trackable point in space. This allows the system to determine the exact trajectory of the club and the ball's position 240 times per second, creating a high-fidelity data stream for real-time adjustments. The nightmare of integration hell Even with perfect hardware, the "integration hell" phase nearly derailed the project. This is the point where software bugs, magnetic interference, and hardware failures converge. One of the most significant issues was Latency Out-of-Phase Oscillations. Because the tracking cameras and the motor controller were running on different schedules, the club would often "over-correct" for a twist that had already happened. Wighton found that his software was reacting to 10-millisecond-old data by moving the motor to a new position, only for the cameras to send another old update, creating a feedback loop of wrongness. Instead of a full rewrite in C++, he implemented a software hack that synced the motor data with the camera data by intentionally lagging the motor's reported position to match the camera's delay. It's a dirty fix, but in the world of DIY tech, if it works, it’s a feature, not a bug. Future of robotic mini golf The current iteration of the Auto-Correcting Mini Golf Club proves the concept, even allowing for blindfolded trick shots and hitting moving targets. Yet, the physics of a real mini golf course—uneven turf, wall bounces, and friction—require even more computational power. Wighton's current physics simulations take upwards of 25 hours to calculate a single complex bank shot. As he moves toward a more optimized version, this project stands as a celebration of the iterative process. It shows that with enough sensors, custom-machined parts, and clever math, you can turn a game of skill into a demonstration of pure engineering dominance.