Results!

[dayalannair]

Triangle Modulation - BW = 240 MHz, Ns = 200, range bin method

Taken on upper uct road, car traveling at a known speed/controlled experiment. Safety based on a min arrival time of 3 seconds, with a max speed of 60 km/h. Speeds tested: 20, 30, 40, 50, 60, 70 km/h. Video and radar output not perfectly synchronised (magic numbers used) due to unknown time required to render colour maps. Safety meter shows 4 sec - TOA. This allows shorter times (less safe) to be shown with a 'higher' colour.

Notice the max distance and how the safety meter changes as the car approaches. Tracking is needed to fill the gaps, and to ensure a smooth PWM signal if this is to be used for indicating safety.

30 km/h plots

60 km/h plots

Note the late detection shown in the second plot below. Should have detected from the max range down. From a second measurement, it seems fast targets are harder to detect at further distances - could be due to weaker signals having shorter integration times.

50 km/h plots

Note how event though the speed slows at the end, the short range means the turn is unsafe.

[dayalannair]

Notes

• The 40 km/h test is very dirty. Investigate. Seems like the difference is some direct sunlight? - confirmed by Google search
• speed fluctuations due to radar and measuring car and wheel speeds
• need to solve few false alarms -> look at spectra
• videos are NOT PERFECTLY synchronised

[dayalannair]

20km/h 2 car

See false alarms in last frame/s.

[dayalannair]

Python vs. MATLAB

Python seems to have better velocity accuracy: