top of page
Above the Clouds

Hands-Free Remote Balance Maze

Nov-Dec 2020

A two-axis balance maze that can be played remotely via hands-free interaction (distance sensing or through tilting a phone). We hope you inspire remote and ambient companionship in the times of the COVID-19. This was a course group project and I was in charge of CAD design, laser-cutting and fabrication, electronics (sensors), and Arduino programming. 

Narrative Description

Our project, the hands-free balance maze, is a mechanical marble maze that allows different remote control methods: via hands-free sensors or even a mobile app. The purpose of the project is to provide entertainment, propose a new method for interaction & control (with each hand or your phone's tilt controlling an axis of movement), and allow remote control of the device via the MQTT server. The game board can be installed anywhere as long as it’s laying on a flat surface; like a desk or floor.


The maze lies flat with the marble at the start position when no player is present. A player can start the game by simply placing his or her hands over the sensors. The proximity between each hand and each sensor determines the degree of tilting along the axis. The player then needs to get the marble ball to the endpoint while avoiding holes on the board that the ball can fall into. When the ball successfully reaches the end there might be a light or music component signaling the success. Whomever the audience is, they will enjoy the tricky challenge and remember the celebration at the end.

The project will be successful if the device can provide an entertaining experience for the user and that the user feels comfortable controlling the board remotely using the sensors or through a phone. User interaction is key.

Technical Outline


We are planning to make the game board using three layers of plywood boxes connected together using shoulder screws & ball bearings; very much like a Gimbal frame. Two hobby servos will be placed in between every two layers to enable rotation/tilting of the first and second inner boxes. These two servos are connected to two sonar rangers that convert input from different hands into tilting movement along the two axes. The innermost box will contain an empty bottom compartment, which will serve as the marble drop. Anytime the player's marble drops through a hole on the maze surface, they will be able to tilt the board to one side and retrieve it easily.

The maze frame can be made lightweight by using press-fit 6mm laser cut plywood parts, which will enable relatively easy movement of the boxes by the two rotating servos. Shoulder screws and ball bearings will allow the boxes to rotate on their axes. Finally, two sonar rangers will be used as input sensors prompting servo movement, while a phototransistor can be used to sense when a marble reaches the maze finish.


Sketch of the maze box design. Front view (left) and side view (right).

CAD Design

CAD designs of the maze box. The outside view (left); the inside rotating frames and the tilting plane (right).

Prototype Development


In the first phase of prototype development, we mainly assembled the laser cut boards and hobby servos. We made sure that the connecting parts (connected by shoulder screws and bush bearings) between different layers are lightweight enough so that the rotation of the frame can be supported by hobby servos. In this demo, we used two range sensors, with each directly sensing distance to one hand then parsing the data to control the servo. During this phase, the maze requires two hands to control.




Then we mounted the maze board onto the frame and assembled a small box for receiving the ball. We explored the possibility of using only one hand (one water bottle in this demo) to control the board, hoping that such correspondence would be more intuitive for the player. We successfully constructed a taped play area in which diagonally placed range sensors can receive accurate x and y data of the hand's position. We applied filters and tested transmitting the data across the MQTT bridge so we can achieve remote playing. We did successfully play the game remotely across half the earth. However, we also noticed that there is a significant lag between the movement of the water bottle and corresponding changes in the data received. The lag results in a great challenge in providing a smooth playing experience for the player.

bottom of page