Radar in Self-Driving Cars
What is Radar?
Radar stands for Radio Detection And Ranging. It has been used since 1999 in vehicles to support drivers with their driving tasks. Today’s radar isn’t very different from the low-resolution radar used in the older systems in terms of the underlying technology. Although few companies have developed newer versions with more resolution and the ability to distinguish between different objects which closes the gap to other sensors widely used in autonomous cars such as lidar and camera. Some applications of radar are in adaptive cruise control (ACC), predictive emergency braking (PEBS) and blind-spot monitoring (BSM) systems.
Physics of Radar:
Radar radio frequency (RF) generator emits electromagnetic waves and its receiver picks up, filters and processes the waves rebounded from the metal surfaces or other reflecting materials. The distance to the objects is measured by multiplying speed of the radio wave (speed of light~300,000 km/s) by half of the propagation time (transmission time + receiving time). Radar can also measure the relative speed of all the detected objects using the doppler shift of the reflected electromagnetic wave as well as the transverse offset with an angle estimation.
Types of Radar:
There are different types of radars based on the range and beam angle. In the past the main types were short-range radars (SRR) in the ~24 GHz frequency band and long-range radars (LRR) in the ~76 GHz frequency band. But the applications are merging and some radars cover both ranges known as mid-range radars (MRR) in the 76 to 81 GHz. Eventually MRRs are expected to dominate the market supplemented by a LRR. Short-range radars typically have a range of 20–50 meters and a beam angle of up to 160 degrees. Long-range radars have ranges of 250 m and beam angles of up to 30 degrees. Mid-range radars have ranges of 100–150m and a beam angle in between.
What are their strengths?
- They’re able to see through objects unlike lidar.
- They work normally in bad weather (rain, snow, dust) unlike lidar.
- They’re relatively cheap and inexpensive.
- They work well in fog and low light night time situations unlike cameras.
- Some newer versions have resolutions and object recognition capabilities comparable to lidar.
What are their weaknesses?
- They don’t have the resolution to detect small objects or multiple objects moving at fast speeds.
- Shorter field of view and accuracy compared to lidar.
- They can’t see color.
Who are the players?
There are many companies involved in developing radar sensor technology such as:
- MAGNA
- Denso
- Bosch
- Valeo
- Autoliv
- Aptiv
- Continental
- Metawave
- Echodyne
- and more …
Future of Radars:
Future objective of research and development for this sensor can be summarized into:
- Increased resolution and wider field of view.
- Long range 3D scanning
- Object detection, classification and tracking.
References:
Bosch, MAGNA, Velodyne, Metawave, Voyage
