Tiny Claw is a desktop claw machine designed to positively reshape how people think about snacking and exercise. Dieting can often feel punishing — especially when it comes to snacking, which many people associate with guilt or failure. Our goal was to turn this idea on its head. Instead of snacks being a setback, what if they became a reward?
Tiny Claw gives users a chance to win a sweet treat — but only if they meet their health goals. For example, walking 5000 steps in a day earns you an attempt on the claw machine. This approach aims to create a more positive, motivational relationship with snacking and fitness.
How does it work?Being a desktop machine, Tiny Claw was designed with space efficiency as the main priority. We made the following major design choices with this in mind:
- CoreXY Mechanism
This 2 DOF mechanism allows for smooth movement in both X and Y directions while keeping both the motors stationary, minimizing space and reducing motor size since the moving claw is light and low-inertia. The XY motors are fixed under the sweet area, alongside the electronics, and each drive a shaft that is connected to a driving pulley. An intricate series of idler pulleys then guide a cable through the central platform, enabling 2 DOF movement of the claw in the XY plane. By synchronising the motion of the 2 motors, the X and Y motion of the claw can be controlled independently - Cable-Based Power Transmission
Instead of belts or gears, we used cable to transmit power. This drastically reduced the size of the pulleys used in the CoreXY mechanism and reducing the overall mechanical footprint of the device
User Interface & Electronics:
The user operates the claw using a joystick and a button on the front panel. Movement is controlled by a Particle Photon 2, which interprets joystick input and handles the logic for the pick-up sequence. Here's how it works:
- The user moves the claw above the desired sweets with the joystick
- Pressing the button lowers the claw, closes it, and lifts it back up
- If a sweet is successfully grabbed, the user can move the claw over the drop zone and release it by pressing the button again
Health Goal Integration:
Attempts are not unlimited — they are earned. A Python script regularly checks a user's Google Fit data via the API. If the user walks more than 5000 steps in a day, the script calls an endpoint on the Particle Photon API to increment the available attempts for that user.
CoreXY Design
The majority of design iterations on our XY mechanism were focused on maximising the range of motion in each axis and minimising friction, the number of components, and the overall size of the mechanism
Gripper Design
The gripper mechanism underwent several iterations as we refined both its reliability and space efficiency. Our initial design aimed to minimise components and maximise space efficiency by coupling the claw's vertical (Z-axis) movement with its opening and closing motion, allowing both actions to be driven by a single motor. A solenoid acted as a latch to hold the claw at the top after attempting to grab a sweet.
While this approach helped reduce the number of motors and saved space, it proved unreliable in practice — the complexity of coordinating both actions with one motor led to inconsistent operation and, in most cases, led to failure to grip sweets correctly. As a result, we decoupled the two motions, assigning one motor to control the up-and-down movement of the claw and another to control the gripper's open/close action. This significantly improved reliability and simplified the control logic.
The gripper opening/closing mechanism itself then also went through two designs iterations - (1) A worm and pinion gear mechanism and (2) A compact gearbox repurposed from the original motor setup. The second design proved far more consistent and robust, ultimately becoming the final choice for Tiny Claw.
Housing & Electronics Enclosure
Initially, we built a rough prototype housing using cardboard to validate our core mechanisms, especially the XY movement and claw actuation system. This version helped us quickly test both mechanical and electronics design, make adjustments, and identify what worked. For testing, we added a breadboard under the housing with all the components from the circuit diagram.
Once we were confident in the mechanics, we moved to a laser-cut housing with MDF, which offered more precision, better stability, and a clean, finished look suitable for desktop use. This version also allowed us to optimise internal space usage, improve cable routing and allowed us to create space for housing the electronics underneath the sweet area.
Comments