Cars see people with enhanced radar

Concepts for boosting WiFi coverage could also help car sensors distinguish walking pedestrians from other slowly moving objects.

Automotive radar can fall short is resolution which can lead to confusion in distinguishing between walking pedestrians and slow-moving vehicles. This is where multiple-input multiple-output (MIMO) concepts fits in.

Multiple-input multiple-output (MIMO) concepts that have improved WiFi coverage around homes and offices could also benefit automotive radar, as demonstrated by researchers in Malaysia and Japan.

Instead of using a single emitter and collector to send and receive information, MIMO-based designs feature an array of transmitters and detectors. The configuration generates a much richer data set of digital signals for computers to process.

So how does this help vehicle sensing? Radar can travel around objects that would ordinarily block a driver’s line of sight. It is also much less affected by bad weather, such as fog, compared to camera-based systems. But one area where automotive radar can fall short is resolution, which can lead to confusion in distinguishing between walking pedestrians and slow-moving vehicles. This is where MIMO fits in.

Scientists from Universiti Teknologi MARA in Malaysia and Tokyo Denki University in Japan compared the performance of a single input single output (SISO) Doppler radar with a MIMObased radar. Both types detected moving targets by sensing a Doppler shift in the frequency of emitted pulses and reflected signals bouncing off objects in motion. Advancing objects produce a positive shift in frequency, and targets heading away from the detector can be identified by negative changes in the reflected signals.

But thanks to the richer data set, MIMO measurements produced much thicker Doppler lines across the recorded spectrogram than SISO measurements.

The MIMO bands contained Doppler shifts, indicating the overall velocity of a target, and so-called micro-Doppler signatures; seen, for example, when walking pedestrians swing their arms. The thicker the line, the easier it is for the radar system to interpret these signatures correctly.

The MIMO configuration provided a clear distinction between different kinds of approaching targets, while there was a degree of overlap in the data during the SISO experiments. The finding indicates MIMO-enhanced Doppler estimation could improve automotive detection schemes. If fitted to a typical car, this enhanced radar system could alert drivers to pedestrians in the process of crossing the road while obscured by parked vehicles.

The researchers plan to develop a working prototype of the MIMO-based automotive radar technology and then use it to take real-time measurements of various obstacles on the road, including pedestrians. Numerous outdoor trials are vital to train computers to distinguish between pedestrians and vehicles so they can accurately alert drivers of regular cars, and even self-driving cars, to make precise decisions.

For further information, contact:
Dr Idnin Pasya | E-mail: [email protected]
Microwave Research Institute
Universiti Teknologi MARA

Published: 11 Feb 2019

Contact details:

Chief Information Officer (CIO)

Institute of Research, Development and Commersialisation (IRDC) Universiti Teknologi MARA (UiTM) Shah Alam, 50450 Shah Alam Selangor Malaysia

News topics: 
Academic discipline: