Intelligent Packing Assistant

Intelligent Packing Assistant is a voice-driven robotic packing prototype built by Team 15 for the Intel Bimanual VLA Manipulation with Multi-Modal Reasoning track. The system demonstrates how a natural-language packing instruction can be transformed into structured and executable robot actions. A user provides a spoken instruction, Speechmatics performs real-time speech-to-text, and the resulting instruction is processed by our Python/FastAPI backend. The backend extracts the relevant packing constraints and generates a structured packing plan. For the MVP scenario, the system works with three object types: Heavy, Normal, and Fragile. The generated plan explicitly assigns robotic actions and arms: the Heavy object uses a coordinated bimanual pick-and-place action with arm1 and arm2, while Normal and Fragile are handled by individual arms. The structured plan is exposed through a deployed FastAPI service on Render and connected to a Three.js simulation. The simulator visualizes the robot execution and demonstrates the coordinated two-arm manipulation required by the challenge. The demonstrated workflow is: Voice Command → Speech-to-Text → Instruction Understanding → Packing Plan → Robot Actions → Bimanual Execution → Final Placement The project focuses on building a practical end-to-end prototype that combines voice interaction, structured task planning, backend APIs, and robotic simulation within a modern AI infrastructure workflow.