A Range-Doppler Map (RDM) is a 2D representation of radar data that encodes the range and velocity of detected targets. It is generated by analyzing the frequency and phase shifts in the reflected radar signals. The horizontal axis corresponds to range (distance from the radar), while the vertical axis corresponds to Doppler velocity (relative radial speed).
HardwarePSOC6 AI kit contains BGT60TR13C XENSIV™ 60GHz radar containing 1 Transmitter and 3 receivers.
PrerequisitesBefore beginning the process, ensure the streaming protocol is installed and running on your microcontroller. Please follow the tutorial for guidance.
Steps- Create a new project in DeepCraft.
- Add Radar node and select Micro Presence as "Use Case".
- The radar input is of shape 1x16x128 from 1 receiver, 16 chirps per frame, and 128 samples per chirp.
- First, we would remove the DC component from the chirps by subtracting the mean across axis 0.
- Next, we will compute the "Real Discrete Fourier Transform" along the range axis after applying the "hamming smoothing" for better detection.
Since the previous transform gives us results in complex form, we can apply the "Complex Discrete Fourier Transform" along the velocity axis (the slow axis). Additionally, we may use "Hamming smoothing" and apply an "FFT shift" to center the zero velocity.
- Now calculate the power and convert to decibels reshape and visualize, you can optionally scale and shift for better visualization.
- We can also use the "Gesture" use case and get the data from all 3 receivers with slight modifications
- Different receivers will have varying gains due to slight changes in angles to the obstacle, resulting in dots of different colors as we utilize RGB channels, with one channel assigned to each receiver.
I am currently unable to remove the artefact on top of the RD plot.
Comments