We were inspired by the idea of blending robotics, AI, and interactive systems to create something both fun and technically challenging. The concept of a smart Nerf gun turret came from a mix of sci-fi influences and a desire to apply our skills in real-time systems, machine learning, and embedded programming in a creative way.

Project Artemis is an autonomous Nerf gun turret that can detect and identify human faces using a custom-trained face recognition model. It features:

• Voice command activation via the Google Voice Recognition API
• Manual control with a PS4 controller using the SDL2 library
• Serial communication to interact with hardware components
• Patrol mode, where the turret scans its environment for targets
• Autonomous firing when a recognized face or target is detected

We developed Project Artemis using:

• C++ as the main programming language
• CMake for project configuration and building
• vcpkg to manage C++ libraries
• Custom face recognition model, trained using OpenCV
• Google Speech-to-Text API for voice control
• SDL2 for PS4 controller input
• Serial communication libraries to interact with the turret's hardware

We tied everything together into a multi-threaded system that handles input, processing, and actuation in real-time.

• Integrating multiple subsystems (voice, vision, control, and hardware) and managing concurrency
• Training the face recognition model to perform reliably under different lighting and angles
• Tuning the servo controls to move the turret smoothly and accurately
• Serial communication bugs, especially syncing messages across different hardware platforms
• Latency issues in real-time recognition and response

• Building a fully functional AI-powered Nerf turret from scratch
• Successfully integrating face and voice recognition in a real-time C++ system
• Achieving smooth and responsive manual and autonomous turret control
• Creating an immersive experience with patrol mode and intelligent targeting

• Advanced multi-threaded programming and real-time systems design
• How to train and deploy a face recognition model with OpenCV
• Practical integration of software and hardware components
• Debugging and optimizing performance across different subsystems

• Implementing target vs non-target classification to avoid firing on authorized individuals
• Upgrading to a more powerful onboard computer
• Adding web or mobile control interfaces
• Integrating object tracking and predictive targeting
• Improving the voice interface to understand more complex commands

• arduino
• c
• c++
• cmake
• linux
• opencv
• python
• sdl2
• shell