Animatronic Human-like Home assistant (Skully)

The 'Skully' animatronic home assistant is an advanced edge-computing and AI-driven system built around a distributed architecture using an ESP32-S3 (Adafruit Memento Camera Board), an ESP32 DevKit v1, and a central processing unit like a Raspberry Pi 5 or laptop. Visual input is captured via the Memento camera board, where Python and OpenCV handle computer vision tasks such as facial tracking to control eye and head movement. Voice interaction begins with a microphone capturing audio, which is processed locally using Vosk for speech-to-text conversion. The transcribed text is then sent to an LLM via the Groq API to generate a contextual response. Piper TTS converts this text response back into speech. To synchronize the mouth movements with the generated audio, the system drives the hardware actuators in real-time. A PCA9685 16-channel servo controller communicates with the microcontroller over I2C to generate the precise PWM signals required to actuate six MG90S micro servos and four heavy-duty MG995 servos, which control the jaw, eyelids, eyes, and neck. Power distribution is managed by a robust 5V 10A Switch-Mode Power Supply (SMPS) backed by multiple 1000uF capacitors to smooth out voltage spikes and current drops caused by simultaneous servo movements. Communication between the various microcontrollers and the core processing unit is handled via ESP-NOW, serial interfaces, and Wi-Fi, allowing the vision, audio processing, and physical motion sub-systems to operate synchronously.