PatotIA - AI Assistant for calculations and 3D simulation

2-3 weeks

Detailed project description

This project represents a complete robotic assistance system that combines machine learning with robotic engineering. The core of the system is a transformer model (CodeT5) specifically trained to interpret natural language commands and convert them into precise mathematical operations for robotics.
Main features:
- Custom AI engine: Trained transformer model that achieves 92.94% accuracy in command interpretation with small dataset (just over 500 sentences)
- Complete robotic calculations: Direct/inverse kinematics, DH transformation matrices, Jacobian analysis
- Interactive 3D simulations: Real-time visualization of the robotic arm using Plotly
- Modern web interface: Responsive frontend with real-time chat
- Contextual memory: System that remembers parameters from previous commands

Usage

  • 1- Clone GitHub repository and download the "Robot_AI_TrainedModel" Release
  • 2- Save it in Backend/models
  • 3- Navigate to main directory via terminal
  • 4- pip install -r required.txt
  • 5- cd Backend
  • 6- python api.py
  • 7- Open http://127.0.0.1:5000 in browser

Achievements

  • Project with significant potential ahead (will be updated)
  • Fine-tuning transformer models for specific tasks
  • Creation of specialized datasets
  • Natural language processing applied to technical domains
  • Taught me, after several failed attempts with Ollama and FlanT5, the value of choosing the right pre-trained model: only CodeT5 offered the technical precision I needed

Project information

Key aspects

Fine-tuning NLP Transformers Pre-trained AI models Contextual memory CodeT5 FlanT5 Ollama Integrated chat visualization

Duration

2-3 weeks