2.5 DoF Carnival Fishing

Our team built a 2.5-degree-of-freedom gantry that turns the carnival fishing game into a machine: the player steers an electromagnet over a spinning plate of fish with a joystick, dips it to pick a fish up, and drops it in the bucket.

The gantry moves freely in X and Y, which gives two full degrees of freedom. The third axis, Z, only has two positions, up and down: one press of the joystick lowers the magnet, pauses, and lifts it back up. That limited vertical motion is the “half” degree of freedom. A fifth stepper motor spins the plate underneath to make the game harder, and everything runs from one handheld joystick and one control board.

This page follows the project in order: the brief, how it plays, the first linear stage, the CAD models, each part as built, manufacturing, electronics and code, and the final assembly.

In six weeks, each team had to design, build, and integrate a Cartesian motion system that performs a task of its choice.

The player lines the magnet up over a fish, switches the magnet on, dips to pick the fish up, and carries it to the bucket.

Pushing the stick further moves the gantry faster, so small corrections are easy and long moves are quick. While the magnet is dipping, X and Y are locked so the player can’t drag the magnet sideways through the fish.

The plate game gets harder each round, with a 3-second pause between rounds:

Before designing the gantry, we built and tested a single linear stage.

The first stage used a NEMA 17 stepper turning a lead screw through a flexible coupler. A brass nut drove a foam-board carriage along the screw, with a pillow-block bearing supporting the far end. We drove it with G-code through Marlin firmware, configured at 400 steps/mm for the 8 mm-per-turn lead screw, then stacked two stages into the in-class plotter.

For the fishing game, we kept the NEMA 17 motors and the MKS Base board but switched to timing belts running on 80/20 aluminum extrusion. Belts move a carriage much faster than a lead screw, which a game needs, and the extrusion serves as both the frame and the rail.

The whole machine was modeled as one SolidWorks assembly, built up from subassemblies for each axis, the plate, and the controls.

We imported vendor models of the NEMA 17 motors, timing pulleys, MKS Base board, and joystick module to check fit and clearances. The motor mounts are driven by global equations (a 1.7 in motor face, 1.2205 in between mounting holes, a 5 mm shaft, and the 1 in extrusion size), so changing one number updates every mount.

Each custom part was designed around the 1 in extrusion, the NEMA 17 motors, or the game itself.

Custom parts were 3D printed; the structure uses stock 80/20 extrusion, timing belts, and off-the-shelf pulleys, all joined with screws.

One MKS Base board (an ATmega2560 running Arduino code) reads the joystick and drives five steppers and the magnet.

We first planned to run the gantry on GRBL with G-code jog commands streamed from a laptop script that read the joystick. We switched to our own Arduino firmware so the joystick plugs straight into the board and no computer is needed during play. The firmware grew one feature at a time:

How the final firmware works:

The core of the joystick control:

// Stick reading -> signed target speed, with dead zone and proportional control
float stickToSpeed(int raw, int center) {
  int d = raw - center;
  if (abs(d) < DEADZONE) return 0;
  float span = (d > 0 ? 1023 - center : center) - DEADZONE;
  float f = (abs(d) - DEADZONE) / span;
  if (f > 1) f = 1;
  return (d > 0 ? f : -f) * MAX_SPEED_XY;
}

// Move current speed toward target, limited by acceleration
float ramp(float current, float target, float dt) {
  float maxDelta = ACCEL_XY * dt;
  if (target > current + maxDelta) return current + maxDelta;
  if (target < current - maxDelta) return current - maxDelta;
  return target;
}

Download the full Arduino sketch (joystick_gantry_final.ino)

The finished gantry stands on 80/20 legs over the fish plate, with the catch bucket beside it and the controls on a cable in front.

At our final demonstration on October 21, our professor took the joystick and played the game, steering the magnet over the spinning plate.