Radar Data Cube Processing and Linux-Based Acquisition Pipeline for TI Cascaded Radar (MMWCAS-RF-EVM)

Texas Instruments’ cascaded radar platform MMWCAS-RF-EVM [1] is a high-performance mmWave sensing system widely used for research in automotive, robotics, and industrial perception. Its multi-chip configuration enables large virtual antenna arrays and high-resolution 3D processing through radar data cubes. While TI provides a complete reference pipeline for Windows, many research and production environments require Linux-based operation and the ability to automate data acquisition without relying on TI’s proprietary GUI tools. This is also the case of the field robotics research group at Örebro University. 

The goal of this thesis is to study, implement, and evaluate radar data cube processing workflows for the TI cascaded radar. The student will first set up the official TI pipeline [1], record data, and visualize the generated radar cubes. A key step is to perform an experiment using a radar reflector (with known Radar Cross Section, RCS) to validate and calibrate the radar measurements. The second phase of the work focuses on creating a Linux-compatible pipeline that configures and starts radar measurements from Linux, while allowing the device to continue storing raw data on the onboard SSD — removing the dependency on the TI Windows software and enabling integration into robotics/automation pipelines. The pipeline can draw inspiration from open-source projects available at GitHub [2, 3, 4]

This thesis will be supervised by Örebro University, namely by the researchers at the Robot Navigation and Perception laboratory. Deliverables include documented experiments for the reference pipeline, documented code for the custom pipelines, a final thesis report, and a live demonstration of the Linux-based measurement workflow.

[1] AWRx Cascaded Radar RF Evaluation Module (MMWCAS-RF-EVM) User's Guide https://www.ti.com/lit/pdf/swru553

[2] https://github.com/azinke/mmwave

[3] https://github.com/petersvenningsson/MMWCAS-RF-EVM-SP-chain

[4] https://github.com/RaDelft/RaDelft-Dataset

Project Objectives

• Set up and execute the official TI radar data cube processing pipeline for MMWCAS-RF-EVM.
• Conduct a controlled experiment using a radar reflector with known RCS.
• Record raw radar data, generate Range-Doppler-Angle radar cubes, and visualize them.
• Calibrate the radar reflector response and estimate or validate its effective RCS in your setup.
• Design and implement a Linux-based radar activation, data recording and processing pipeline, enabling:
  • Switching the radar on/off from Linux via UART/Ethernet.
  • Triggering measurement sessions without TI GUI tools.
  • Ensuring all raw data is still stored on the onboard SSD.
  • Processing the stored radar data into the radar data cube format (Range-Doppler-Angles).

Who Should Apply

This project is suitable for students interested in radar signal processing, robotics perception, or embedded sensing systems. You should enjoy working with real hardware and interpreting high-dimensional sensor data.

Required qualifications:

• Good programming skills in Python and/or C++,
• Basic understanding of signal processing, linear algebra, and frequency-modulated radar systems.

Recommended qualifications:

• Experience with mmWave radar, ROS/ROS 2, embedded Linux, or real-time sensor pipelines,
• Background in wireless sensing or estimation theory.