IEEE Smart and Radio Controlled Marshmallow Cannon
A smart, RC car featuring a marshmallow cannon powered by AI-driven auto-tracking and RF communication.
Jun 2024 - Aug 2024 (3 months)
Overview
As part of a cross-functional team of engineering students, I contributed to an innovative RC car project built from scratch. The vehicle featured an automated marshmallow launcher controlled by an AI-powered targeting system.
I focused specifically on developing the facial recognition and tracking components that guided the launcher's targeting system. Using Python, OpenCV, and Dlib libraries, I implemented a real-time detection system that controlled the aiming mechanism.
The solution combined Haar Cascade classifiers for initial face detection with KCF (Kernelized Correlation Filters) tracking algorithms, chosen for their efficiency and low computational overhead. This implementation enabled precise targeting while maintaining responsive performance.
Challenges
- •Finding a model that balanced accuracy and speed for real-time tracking on a Raspberry Pi
- •Mapping the tracking data to the turret's movement for accurate targeting
- •Integrating the tracking system with the existing RC car controls