How yorick-xy/destis works
Project Type and Purpose
• Destis is a hybrid robotics + home-automation application with an AI/ML component. It is neither a classic web full-stack app nor a pure library: it integrates
• an autonomous robot (navigation, obstacle avoidance, motor control, sensors)
• a home-automation subsystem (device control via MQTT, voice commands)
• AI/ML modules (chatbot, NLP, vision, speech)
• a CLI interface (src/main.py) for orchestration and user interaction
File‐Structure Analysis
• Top‐level scripts & configs:
• setup.py, requirements.txt – packaging and dependencies
• scripts/ – deploy.sh, format_code.sh, run_tests.sh
• data/ – potential datasets or simulation logs
• architecture.txt – placeholder for existing architectural notes
• Source tree (src/): four main packages:
• ai/ – chatbot.py, core.py, models.py, NLP, speech_recognition/synthesis, vision, etc.
• home_automation/ – mqtt_client.py, device_control.py, routines.py
• robot/ – sensors.py, navigation.py, obstacle_avoidance.py, motor_control.py, mock_gpio.py
• utils/ – config readers (YAML), database, logger, exception_handler
• Entry point: src/main.py ties everything together
• tests/ – unit tests for each subsystem
README Insights
• Features: autonomous navigation, obstacle avoidance, device control via voice/MQTT, AI conversational engine
• Simulation constraints: hardware calls are mocked, only chatbot is fully functional
• Relies on DialoGPT (via PyTorch) for conversation
a. Main Components
• User Interface (CLI)
• Orchestrator (main.py)
• AI Subsystem
– Data Preprocessing
– Models (DialoGPT, vision, speech)
– NLP, SpeechRecognition, SpeechSynthesis
• Home Automation Subsystem
– Routines Engine
– MQTT Client
– Device Control
• Robot Control Subsystem
– Sensor Interface
– Obstacle Avoidance
– Navigation Planner
– Motor Control / GPIO (mocked)
• Utility Services
– Configuration Loader (YAML)
– Logging & Exception Handling
– Database (local state, logs)
• External Services & Hardware
– MQTT Broker
– Sensors & Actuators (motors, distance sensors)
– (Optionally) DialoGPT Cloud or Local Model
b. Relationships & Data Flows
1. User issues a command at the CLI → main.py
2. main.py parses input → dispatches to AI Subsystem (for voice/NLP) or to Control Subsystems
3. AI Subsystem may call Data Preprocessing → Models → returns intent/response
4. If home-automation intent: orchestrator → home_automation.routines → mqtt_client → external MQTT Broker → devices
5. If robot intent: orchestrator → robot.sensors (reads) → obstacle_avoidance → navigation → motor_control → actuators
6. Throughout, utils/logger logs events; utils/database persists state; utils/config supplies parameters
c. Architectural Patterns & Principles
• Modular layering: clear separation between AI, control, utils
• Dependency injection via utils/config
• Mocking hardware in tests (mock_gpio) for CI
• Single entry point (main.py) orchestrates modules
d. Key Technologies & Frameworks
• Python 3.7+
• PyTorch / HuggingFace (DialoGPT)
• paho-mqtt (or similar) for MQTT client
• YAML for configuration
• unittest/pytest for tests
Final Advice
• Incrementally refine: add utils/config & database as supporting services