Client Meeting Minutes

Date: October 9, 2024
Time: 10:00 AM - 11:00 AM

Action Items

1. Recurring Calls
   • Setup recurring calls until November 14. (Assigned to GP)
2. Overview Slides
   • Send slides that briefly show the flow and required software. (Assigned to GP)
   • Let me know if any clarification is needed on the attached document.
3. Contracts
   • Contracts need to be reviewed by TI legal department. (Assigned to GP)
   • Follow-up on status by October 16.
4. Board Shipment
   • Send boards to Faik by tomorrow. (Assigned to GP)
5. File Sharing
   • Clarify if there is a file-sharing site already in place to start sharing files.
6. Minimum Viable Product (MVP) for November 14
   • Define the MVP scope and feasibility for the upcoming deadline.
7. Team Structure
   • Discuss the division of roles and leadership within the team.

Stakeholders

• The radar technology aims to offer privacy advantages over camera-based systems by providing improved privacy protections.
• Potential Use Cases:
  • Healthcare:
    • Monitoring patient movements in hospital rooms, such as detecting whether a patient is in bed, standing, or has fallen.
    • Enabling remote monitoring of patients at home to measure activity and prevent accidents.
  • Public Transportation:
    • Detecting the position of individuals on train platforms, including their movements and potential risks.

End-Game Requirements

MVP Requirements (will not be done by the ddl):

• Detect a person’s or group’s positions (e.g., lying, sitting, standing) and intermediate positions. Optionally include walking and running detection.
• Utilize a point cloud provided by the radar. Machine learning analysis will run on a separate PC.

Development Phases:

1. Part 0.5: Understand the workflow of the radar board.
2. Part 1: Recreate results using the existing workflow.
3. Part 1.5: Analyze data patterns to understand what the data represents.
4. Part 2: Apply machine learning to test people detection, compile the model, and generate the flashable binary for the radar board.

Technical Requirements:

• Embedded C: Will be used for radar firmware development.
• Python: Will be used for data analysis.
• Radar Flow: The process depends on the application, with code for embedded C being provided.
• Board Preparation: Boards need to be flashed before data production; they will not arrive data-ready.
• Machine Learning Model: A preliminary PyTorch program exists for the project.

Software Requirements:

• The software list will be provided later.
• Linux Environment: Required for model conversion (.onnx to C code) using TVM compiled for Ubuntu.