As described in Figure 2-1, the implementation of the obstacle detection application example in the
signal-processing chain consists of the following blocks implemented as DSP code
executing on the C674x core in the AWR1843AOP.
- Range processing
- For each antenna, 1D
windowing, and 1D fast Fourier Transform (FFT)
- Range processing is
interleaved with the active chirp time of the frame
- Range-angle heat-map generation
- Generate angle spectrum for
each range bin using covariance beamforming or Capon beamforming.
- Outputs Range-Elevation
angle heat-map and Range-Azimuth angle heat-map.
- Object Detection
- On each range-angle
heat-map, search a single peak cross angle for each range bin, then
apply one dimension CFAR check on the peak angle cross the neighboring
range bins.
- Take the union (or
interception) of the two peak sets detected from the two range-angle
heat-maps.
- Doppler estimation
- For each detected range
bin, estimate the Doppler output and apply non-coherent combination
among all antennas. Find the peak index and used it as the Doppler index
of the detected target.
- Outputs the Range-Doppler
output of this Doppler peak index and used later for angle
estimation.
- Angle of arrival estimation
- Calculate the
two-dimensional angle spectrum using 2D FFT.
- Single peak search on this
two-dimensional angle spectrum and calculate azimuth and elevation angle
associated with the peak location.
- Clustering
- Perform DBSCAN-based
clustering algorithm every fixed number of frames and report the
results.
- Output the number of
clustering and properties like clustering center and size.