Date: 24-10-2025
Time: 9:00 - 10:00
Team: Alina Zubova, Dan Robb, Henry Edwards, Talal Aljallal, Tom Lam, Michal Berkasiuk

Key Project Learnings & Context​

1. Filter Gap (Last Year): A critical lesson from the previous project was the absence of a final filter on the results. This suggests a major area for us to improve upon in our pipeline.
2. AppImage Binaries: These are essentially the compiled code for the radar chip. Their job is to convert the raw radar information into the specific data format we need for processing (e.g., point cloud).
3. Chirp Configuration: This refers to the specific kind of chirps (radar signals) the hardware sends out, which directly affects measurements like range. Good news: it's built into the chip, so we shouldn't need to worry about configuring it ourselves.

Minimal Viable Product (MVP) Focus​

The MVP goal for this term is two-fold:

1. Replication + Improvement: We need to reproduce the previous team's output but rely only on the radar unit for the data source.
2. ML Demo: Get a basic demonstration working that involves training a model. This means we need to immediately focus on dataset handling and collection.

ML System Requirements​

• The trained model should be running locally on our machines.
• The model must be built using the PyTorch framework.

Data Collection & Visualisation Strategy​

Data Types​

The raw data we collect will be a mix of:

• Point Cloud Data: A set of 3D points representing detected objects.
• Tracks Data: Time-series data tracking the movement of detected objects.

Visualiser Goal​

If we develop a visualiser, a high-value feature would be to superimpose the radar information (e.g., detected points/boxes) directly onto the live video feed for immediate context and verification.

Recording Specifics​

1. Radar Height: Keep the radar unit fairly low during recording, roughly 1m off the ground.
2. Movement: When collecting data, ensure people walk right-to-left AND left-to-right in front of the radar.
3. Final Classification Filter: We can use data like height and speed as a crucial final filter to help classify or distinguish different objects/people.

Date: 17-10-2025
Time: 11:00 - 11:35
Team: Alina Zubova, Dan Robb, Henry Edwards, Talal Aljallal, Tom Lam, Michal Berkasiuk

Key Discussion Points​

Project Resources & Equipment​

• Radar Inventory: Confirmed we currently have four radar units.
• Acquisition: We can secure one additional radar from last year's team. We will also follow up on a potential unit from Sarah.

Technical & Setup Status​

• Industrial Visualizer: Assisted Dan with successfully getting the visualizer application running.
• Last Year's Demo: This week’s core technical task is to get the demo application from the previous year’s team running on our local development machines.

Documentation & Organisation​

• Meeting Notes Structure: Discussed separating non-logistical, high-level project information from the regular meeting notes and storing it in a dedicated folder for better organization.

Action Items & Next Steps​

Date: 17-10-2025
Time: 11:00 - 11:35
Mentor: Jack Cains
Team: Alina Zubova, Dan Robb, Henry Edwards, Talal Aljallal, Tom Lam

Next Steps & Hardware​

1. Run Demos on Boards: The immediate next step is to start running the provided demos on the radar hardware boards.
2. Stakeholder Ideas: Remember to ask client for suggestions on potential stakeholders. Good to have a wider perspective on who might use our work.
3. Hardware Insight: The team last year 3D printed a holder for their board. Worth keeping in mind for managing the hardware setup.

Kanban & Branch Workflow​

Branch Naming Convention​

Adopt this structure for all new branches:

category/issue-number/short-description

• Example: docs/#55/update-readme

General Workflow​

1. Branch Cleanup: It's a good practice to delete branches after they are merged to keep the repository tidy.
2. Kanban Discipline: Ensure that both Issues and Pull Requests (PRs) clearly reflect their current status on the Kanban board (e.g., To Do, In Progress, Review).
3. Issue Detail: Make sure all issues have a clear description and a milestone assigned. Even for docs and meting notes - just good practice

Date: 10-10-2025
Time: 09:00-10:00
Team: Alina Zubova, Dan Robb, Henry Edwards, Talal Aljallal, Tom Lam

Key Decisions and Discussion Points​

1. Hardware and Radar Model​

• The team will begin development using the IWRL6432 radar model.
• The project may potentially pivot to the 6864 model later if necessary.
• Greg will be sending Dan the required hardware.

2. Meetings and Communication​

• Weekly team/client meetings will be held at 9:00 AM.
• All meetings will be conducted online.
• Next week's meeting (beginning 13/10/2025) is cancelled/not needed..
• Action: The team must sort out a secure way to share files with Greg (e.g., shared cloud folder, dedicated GitHub repository).

3. Data Collection Planning​

• Look into getting a camera, tripod, and webcam.
• Potential location for initial data collection: Queen's Lecture Theatre.
• Action: The team needs to think about booking a room suitable for data collection and testing.

4. Technical Development Setup​

• Development Environment Setup: Get the development environment configured, including Jupyter, VS Code, and PyTorch.
• Resources: Use TI's official documentation or last year's team documentation for setup guidance.
• Initial Model: The team will start with a simple 3-layer fully connected model as a learning exercise.
• Future Model: Plan to transition to a 1D CNN model (1-Dimensional Convolutional Neural Network) for future development.

Next Steps / To-Do​

1. Dan: Await receipt of radar hardware from Greg.
2. Team: Set up the development environment (Jupyter, VS Code, PyTorch).
3. Team: Determine and establish a secure file-sharing method with Greg.
4. Team: Investigate and book a room for data collection (e.g., Queen's Lecture Theatre).

Date: 10-10-2025
Time: 11:00-12:00
Mentor: Jack Cains
Team: Alina Zubova, Dan Robb, Henry Edwards, Talal Aljallal, Tom Lam

Workflow & Git Practices​

• Main Branch Rule: Do not commit or merge directly into the main branch.
• Feature Branching: All development work must be done on a feature branch branched off of dev.
• Branch Naming: Ensure feature branches are given descriptive and useful names (e.g., feature/add-docusaurus-research).
• Commit Consistency (Contribution Tracking): This is about team equity. Make sure that all team members are contributing a similar volume of work (commits or lines of code) to the project.
  • Commits should not be too "uneven" across team members.
  • If contributions become significantly skewed toward one or two members, we will have to justify the workload distribution for the project review.

Hardware Logistics​

• Radar Chip Storage: The radar chips can be stored securely in the Hackspace drawers.
• Camera/Tripod Acquisition: The team can potentially acquire the required camera and tripod from the MVB Tech Office.

Next Steps / To-Do​

1. Team: Ensure all new branches are created from dev.
2. Team: Collect the camera and tripod from the MVB Tech Office.

Date: 09-10-2025
Time: 13:00-14:00
Team: Alina Zubova, Dan Robb, Henry Edwards, Talal Aljallal, Tom Lam

Questions for Greg​

(In order of priority, as discussed)

🧭 Core Questions​

1. Project Brief

   • Could you give a more in-depth explanation of the project brief? (In other words — what exactly are we meant to be doing?)

2. Contracts / Licenses

   • We’ll likely send over the license options or a short document explaining them for you to review.

3. Radar Hardware

   • When will we receive the radar, and which model will it be?

4. Next Steps

   • What should our next step be?

     • More research?
     • Or can we begin looking at some code?

5. Example Data / Replay

   • Is it possible to get example data or a “replay” mode that simulates the radar output without having the hardware?

6. Multiple Chips

   • Would we be able to get access to multiple radar chips?

7. Future Meetings

   • Should we schedule follow-up meetings now?
   • Would you like us to prepare an agenda for future meetings? (I can handle this — no worries!)

💡 Additional Asides​

• Anything we can start coding without the radar hardware?
• Clarify the C and binary files mentioned by Alina Zubova — not entirely sure what that referred to.
• Will you want more release stages beyond MVP → Beta → Final?
• Confirm the tech stack — what languages and tools will we actually be coding in (though we may already have an idea).

Next Meetings:

• Team meeting: 9:00 10/10
• Meeting with Jack: 11:15 10/10

Date: 07-10-2025
Time: 15:00-17:00
Team: Alina Zubova, Dan Robb, Henry Edwards, Talal Aljallal, Tom Lam

During this meeting, the team focused on aligning upcoming tasks and ensuring everyone understood their individual responsibilities. We discussed progress on the ethics pre-approval form and clarified what still needs to be completed before submission. The team also revisited the different contract options, confirming that we’ll continue evaluating which license type fits best for our project.

We agreed to continue researching the technical aspects of FMCW radar and to go through the documentation provided by Texas Instruments in more detail. The goal is to identify key examples, tutorials, or reference materials that will help us understand the hardware and software workflow more effectively. The team also discussed the tech stack and tools that Texas Instruments expects us to use which includes primarily working with Python, Pytorch, Code Compressor Studio (CCS), and Jupyter Notebooks.

The meeting helped clarify immediate priorities and ensured that everyone knows their responsibilities for the week. For our next meeting which is scheduled on the 9th of October 2025. We will discuss the questions we will need to ask the client on the first client meeting.

Date: 06-10-2025
Time: 14:00-15:00
Team: Alina Zubova, Dan Robb, Henry Edwards, Talal Aljallal, Tom Lam

Topics Discussed​

• Kanban & issue templates setup.
• IP and contract choice (leaning towards company open source contract).
• Ethics pre-approval form.
• UK GDPR / data handling.
• Pull request template.
• Bringing the dev branch up to date.
• Preparing questions for the client on the 7 Oct 2025 workshop.
• Researching the documentation provided by the client on FMCW Radars.

Actions to Take​

Ask the unit director for guidance on:

1. The Ethics form.
2. Correct contract type.
3. Each team member to go through documentation and research how the project will work.
4. Summarize findings and add research notes to Docusaurus.
5. Set up machines with the required software from Texas Instruments' website.

These discussions focused on aligning the team’s workflow and ensuring administrative steps (contracts, ethics, GDPR) are progressing correctly.
The goal is to establish a solid foundation for the technical phase of the project — setting up tools, clarifying expectations, and improving collaboration through documentation and consistent updates on Docusaurus.

Date: 03-10-2025
Time: 11:00-12:00
Mentor: Jack Cains
Team: Alina Zubova, Dan Robb, Henry Edwards, Talal Aljallal, Tom Lam

Project & Constraints​

1. New Domain & Hardware – The team is entering a new technical space. Expect a learning curve around radar systems and their documentation.
2. Client Responsiveness – Texas Instruments may be slow to reply; in the meantime, focus on improving the README and internal resources.
3. Agile Workflow – Prioritize Agile principles throughout development.

MVP​

A research deliverable that demonstrates understanding of TI FMCW Radar and the overall project.

Possible inclusions:

• A report explaining FMCW principles and TI radar capabilities
• Feasible pathways for modelling or visualization

Tone:
“We’ve never done this before” → Embrace the challenge positively. Highlight the novelty and ambition of the project.

Ways of Working (Agile)​

• Kanban Board: GitHub Projects → Backlog → To Do → In Progress → Done
• Stand-ups: Brief, regular meetings
• Documentation: Shared notes from each client/mentor touchpoint

Questions for the Client (TI)​

1. Tips for newcomers to radar tech; recommended starting points in their docs
2. Key documentation pointers – must-read sections, examples, and known pitfalls
3. Preferred meeting frequency and expected response time for questions
4. Clarify what an MVP looks like for this term (given no ML background)
5. Confirm required technology (Radar hardware) to be used and delivered to the team