How Automotive Radars Are Advancing Safety Features
Have you ever wondered how self-driving cars navigate our chaotic traffic and complicated road infrastructure, things that even many experienced drivers struggle with? Come to think of it, how do our cars know the safest cruising speed on the freeway or how high the curb is when reversing?
In all these situations, automotive radars play a key role. They're also essential in the autonomous vehicles and intelligent transport systems of the future. In this article, learn:
- how they enable driver assistance and autonomous driving
- how they work, in detail
- how to test them
What is an automotive radar?
An automotive radar is a radio detection and ranging (radar) vehicle sensor that uses radio waves to measure the positions and trajectories of vehicles, people, animals, and other objects around it.
It works by transmitting radio waves in the direction of interest. They bounce off objects in the vicinity and return to the sensor. The sensor then compares the characteristics of these reflected waves against the transmitted waves to infer the positions and movements of the objects.
Such automotive radar systems enable superior driver assistance, even autonomous driving, by helping the vehicle remain aware of all the obstacles around it in real time.
An automotive radar can measure these characteristics of other objects:
- distance
- velocity
- direction or bearing
- angular size
What are the different types of automotive radar systems?
Automotive radars are traditionally classified based on their maximum distances. There are no official definitions, but a quasi-official technical report from the European Telecommunications Standards Institute uses the following four definitions:
- Long-range radar (LRR) has a range of 150-250 meters or more and a field of view (FOV) of 20-25 degrees.
- Medium-range radar (MRR) is for a range of 50-150 meters and an FOV of 50-60 degrees.
- Short-range radar (SRR) is for distances of 15-50 meters and an FOV of 80-90 degrees.
- Ultra short-range radar (USRR) is for less than 15 meters of distance and an FOV of 90-110 degrees.
But nowadays, the traditional range-based classification, though relevant, is blurred by a new generation of sophisticated high-performance radars called imaging or 4D imaging radars.
Imaging radars provide a radio photo of the scene that's closer to cameras and lidars (light detection and ranging) than traditional radars. Their construction enables capabilities like:
- high angular resolutions in all ranges, in the order of 0.5 degrees compared to the 5 degrees of traditional radars
- the ability to figure out shapes, features, and orientations
- the ability to distinguish and track different types of objects more accurately
How do automotive radars contribute to advanced driver assistance systems (ADAS) and autonomous driving?
In ADAS and autonomous driving systems, automotive radar sensors serve as the eyes and ears along with camera, lidar, and ultrasonic sensors.
The vehicle's electronic control unit (ECU) uses sensor fusion to combine the data from all sensors and form a coherent, real-time 3D picture of the vehicle's surroundings, enabling it to make decisions related to driving assistance and autonomous driving.
Automotive radars enable the following ADAS capabilities:
- adaptive cruise control (ACC)
- lane change assistance
- collision avoidance
- emergency braking
- blind spot detection
How many automotive radar sensors are installed on a car?
For 360-degree awareness, six to 10 radar sensors are typically placed around a car. Waymo's self-driving taxi had six radar sensors in 2022, while Mercedes-Benz used eight radar sensors in their test vehicles.
In general, a fully equipped car should have:
- a front-scanning LRR
- a front-scanning imaging radar
- four corner MRRs or SRRs
- two side SRRs
- one or two rear MRRs or LRRs
What are the benefits of using automotive radars?
Radar technology is in high demand in the automotive industry because it doesn't have the drawbacks of other popular sensors like cameras. The benefits of automotive radars include:
- All-weather performance: Under adverse conditions like rain, fog, dust, sleet, snow, and sandstorms, radar outperforms lidar, and lidar works better than cameras.
- Direct velocity measurement: The Doppler effect offers a direct, physics-based, reliable way to measure the velocity of an object, unlike camera sensors which rely on inexact methods like optical flow analysis. Lidars can also rely on the Doppler effect, but they don’t work under all weather conditions and are more costly.
- Non-line-of-sight detection: Radar can potentially detect obstacles and hazards around corners, which is very helpful in dense urban environments, especially for large commercial vehicles.
- Affordability: Radar is cheaper than lidar and comparable to camera sensors.
What frequencies do car radars use?
The frequency bands used by automotive radars are decided by regulations and concerns of interference.
Modern radar modules mainly operate in the 76-81 GHz E band. LRRs use the 76-77 GHz band because regulations allow higher equivalent isotropic radiated power (EIRP). More power means more range.
Lower-range sensors use the 77-81 GHz band. Since radar resolution is influenced by the sweep bandwidth, 4 GHz of sweep enables far higher resolutions in this band. ADAS systems in the past operated in the 24 GHz band. However, this band was prohibited for automotive radars in new cars by regulators in 2022 due to risks of interference.
In contrast, the 76-81 GHz band is protected from interference. It’s often called the millimeter-wave (mmWave or mm-wave) band in automotive contexts.
How do automotive radars work to detect objects and obstacles?
Frequency-modulated continuous wave (FMCW) radars are the preferred type used in automotive radar applications. Let’s look at how they work.
At a high level, a radio wave is transmitted, reflected from an object, and the position and motion are measured using the following principles:
- Distance measurement: The difference in time between the transmitted and reflected signals is proportional to the distance.
- Velocity measurement: Due to the Doppler effect, an approaching object results in a reflected wave with a slightly higher frequency than the transmitted wave, while a receding object results in a slightly lower frequency. This frequency offset is proportional to the relative velocity of the object.
- Bearing measurement: The direction of an object is determined by using an array of antennas and analyzing the characteristics of the reflected wave hitting each one of them.
The steps below explain HYPERLINK "https://www.keysight.com/us/en/assets/3120-1410/application-notes/Understanding-FMCW-Automotive-Radar.pdf"how FMCW radar signals work in more detail.
1. Transmit signals as chirps
The radar transmits a frequency-modulated continuous wave signal whose amplitude-time plot looks like this:
The radar signal repeats every few milliseconds (ms) or microseconds (μs). Each repetition is called a chirp and typically lasts 10-40 ms for slow chirps and 10-40 μs for fast chirps.
A chirp begins at a starting frequency. The frequency then linearly increases, or sweeps, to a maximum frequency before falling back down to the starting frequency. The sweep bandwidth is 1 GHz (between 76-77 GHz) for LRRs and 2-4 GHz (between 77-81 GHz) for lower-range sensors as shown in the frequency-time plot below:
The need for frequency modulation will be explained in later steps.
2. Receive reflected signals
The signal hits all the objects within its range and FOV, gets reflected from each one, and arrives back at the sensor at different times with the same waveform as shown below:
3. Convert high frequency to intermediate frequency
Since low-frequency signals in the kilohertz range are easier to process than high frequencies like 76-81 GHz range, the received signals are downconverted to lower intermediate frequency (IF) signals by a mixer component. All the signal processing described below is on the IF signals.
4. Measure distances to objects
The time between transmission and reception of the reflected signal is proportional to the distance between the radar and the object. This is just the basic equation of motion, time = total distance traveled / speed.
If R is the distance to an object, the total distance traveled is 2R at the speed of light, c:
The sensor just has to measure that time difference to get the distance to an object. But measuring the time isn't straightforward. This is why the frequency is modulated.
Imagine that the signal was unmodulated with a constant frequency F. A signal with frequency F goes out at time A, and a signal with the same F is received at time B. We can measure the time difference, but the sameness of the frequency creates several unresolvable ambiguities. Is it really a reflection or the sensor's own outgoing signal? Is it another radar's transmission?
In contrast, when the frequency increases linearly, the frequency at any time acts like a unique timestamp:
Now the time difference can be unambiguously inferred by measuring the frequency difference (the beat frequency) between the two signals:
5. Measure velocities
Due to the Doppler effect, approaching or receding objects result in frequency offsets in the beat frequency:
Since there are two unknowns — the distance R and the relative velocity v — two measurements of the beat frequency are taken to solve for both simultaneously.
6. Measure directions and angular extents
Typically, there are two to three transmit and receive channels each to form a multiple-input multiple-output (MIMO) system. The channels are connected to different antenna arrays that are carefully engineered for precise phase differences and beam steering.
So each received signal is processed from multiple angles. The differences in their timings and phases enable the inference of bearings and angular extents.
What are the components inside a radar module?
A radar module consists of the following components:
- RF front-end: This circuit contains the antennas, voltage-controlled oscillator, phase-locked loop, and radar transceiver integrated circuits (ICs).
- Transceiver ICs: These are the ICs that handle the transmit and receive channels. They were typically built as monolithic microwave integrated circuits (MMICs), but the trend is to use complementary metal-oxide semiconductor (CMOS) technology for its size and cost-effectiveness.
- Radio signal processing circuit: This circuit consists of a radio digital signal processor that contains amplifiers, mixers, and analog-to-digital converters for processing the FMCW and IF signals.
- Main processor: The main processor is a system-on-chip (SoC) that stores and processes the measured data, tracks objects, and sends data to the vehicle's ECU via the controller area network (CAN).
- Enclosure: The enclosure's construction and materials affect the radiation patterns and beam directions.
Additionally, imaging radars are constructed by using a cascade of radar ICs.
Given the future outlook of automotive radars, the construction of radar modules will see difficult challenges going forward.
What is the future outlook for automotive radars?
Let’s look at some trends that call for advanced automotive radar testing by companies.
1. Counter interference solutions
As automotive radars become ubiquitous, it's essential to prevent unintentional and malicious interference that can result in fatal collisions due to false positives (objects that don't exist) and false negatives (failure to detect objects that exist).
Countering radar interference is similar to the problems faced by modern mobile networks. Some solutions include:
- digital modulation techniques like orthogonal frequency-division multiplexing instead of analog modulation
- embedding unique codes in the signals
- dynamically changing timings and frequencies
2. Use of artificial intelligence (AI)
Using AI algorithms like deep neural networks running on AI-capable SoCs, radar modules can identify and track objects better over time and reduce problems like noisy data and duplicate detections.
3. Use in intelligent transportation systems
Autonomous radars are essential in intelligent transportation systems and connected autonomous vehicles where sensor information is exchanged vehicle-to-vehicle (V2V) or vehicle-to-infrastructure (V2I) for cooperative traffic efficiency and road safety. Cellular V2X is another emerging technology that uses mobile 5G/6G networks for such exchange.
Become future-proof with Keysight automotive radar solutions
In this article, we showed how automotive radars work, how they enable assisted and autonomous driving vehicles of the future, and what challenges companies can anticipate going forward.
At Keysight, we enable radar IC companies, sensor module manufacturers, and vehicle makers to rigorously test their products and meet all necessary standards. Our automotive radar products include:
- Radar scene emulation: The AD1012A is like a virtual reality headset that projects scenes and objects as radio waves instead of pixels. It enables the simulation of complex real-world scenarios as inputs for an automotive radar module.
- Radar target simulation: The E8719A simulates targets for testing 4D imaging radars.
- Radar testing: Over-the-air test solutions enable testing the functionality, performance, and antenna radiation patterns in a controlled environment.
Contact us for additional information on all our advanced automotive radar testing products!