Embedded Files
Creative Tech Art Installation
This mechanical art piece serves as a small scale demonstration of what could later be scaled into a larger scale art piece. The prototype consists of nine linear servo motors mounted inside a container, which are aligned in a grid pattern. An art piece is mounted to these servo motors, which moves according to their programmed movement. The servo motors can be independently controlled, or be controlled by sensors that allow the motion to react to the environment.
This mechanical art piece serves as a small scale demonstration of what could later be scaled into a larger scale art piece. The prototype consists of nine linear servo motors mounted inside a container, which are aligned in a grid pattern. An art piece is mounted to these servo motors, which moves according to their programmed movement. The servo motors can be independently controlled, or be controlled by sensors that allow the motion to react to the environment.
Flower Art Piece
This art piece is mounted to the end of the actuators using heat set inserts. The actuators have a threaded head attachment that allows the art piece to be easily attached. The center of the art piece is moved up and down by the actuator. This causes the petals to either gradually fold outward due to gravity or to fold inward due to collision with the "flower pot" cylinder. The flower was 3D printed as one piece, with hinges built in to the print, no assembly required.
Software and Control System
The final project software used the Raspberry Pi 5 and OAK-D Lite camera to control the interactive wall system. Beyond meeting the original functional requirements, the project exceeded expectations by implementing multiple AI-based sensing modes to create the intended “WOW!” factor. Models include: depth-based spatial sensing, a AI Person detection model, and a MediaPipe-based AI Palm Tracking model. Additional software features were also completed based on sponsor request, including a final GUI with all scripts and preprogrammed motion options to make it more “plug and play”
Depth Detection
Software was used to detect the depth of the image, allowing for actuation when objects in frame met a set required depth, within a 9x9 grid
Uses the OAK-D Lite camera for spatial depth sensing
Divides the camera view into a 3×3 grid
Measures distance in each region
Maps detected depth regions to the actuator grid
Triggers motion based on user position
AI Palm Detection & Tracking
The software is able to detect a person's palms, and would actuate the corresponding actuator.
Uses a MediaPipe-based palm detection model
Detects palms in real time
Maps palm locations to the 3×3 actuator grid
Supports interactive hand-based control
Uses filtering to improve detection stability
Person Detection
Another detection method involved using AI to detect when a person was in frame.
Detects people through the OAK-D Lite camera and assigns confidence levels whether if an object is a person.
Triggers different wall responses based on person count
Expands interaction beyond basic motion sensing
Added as an extra interactive sensing mode
Custom GUI
Built in Python Tkinter
Controls the wall from one central interface
Runs preprogrammed motion patterns: Snake, Wave, Ripple, Pulse, and more!
Adjusts ramp/control settings
Starts depth, person, and palm tracking modes
Shows live system status
L-12 R Micro Linear Servo
L-12 R Micro Linear Servo
These linear servos are the core of the design, allowing for smooth, controllable linear movement. Each servo has a 100 mm stroke length, a 50:1 gear ratio, and a no load speed of 25 mm/s
Page updated
Report abuse