PiDog is a 7-inch robotic dog built as a personal companion robot. It listens, speaks, and responds to human interaction using open-source AI models. Seeing my younger sister spend hours interacting with PiDog showed me that even limited robots can have real human impact.
- Voice interaction (Speech-to-Text + Text-to-Speech)
- AI conversation using open-source LLMs
- Morse Code Translation
- Raspberry Pi 4
- Sunfounder Kit
- Microphone
- Speaker
- LED Strip
- Operating System: Linux (Raspberry Pi OS)
- Languages: Python
- AI Models: Open-source LLMs (GPT-OSS 120B, Nvidia Nemotron)
- Speech to Text: Vosk
- Text to Speech: gTTS
- Audio Capture: PyAudio
Focuses on companionship, not productivity
Built within student-level constraints
Combines AI, robotics, and emotional design
Designed and integrated end-to-end by one person
Limited mobility due to size Processing constrained by embedded hardware
These limitations shaped the design and helped me prioritize interaction over movement.
Expanded emotional responses
Improved speech recognition accuracy
Camera-based perception
How software decisions affect physical behavior
The challenges of running AI on constrained hardware
Human-robot interaction is as much about design as code
Building something “imperfect but real” teaches more than simulations
Built independently as a student project
Not a commercial product
Open to iteration and experimentation