Have you ever wondered why autonomous vehicles need so many sensors? Driverless cars carry sensors on the roof, windshield, bumpers, and even side mirrors.
Sensors are the eyes of an autonomous car. They allow it to see the world and safely operate within it. But with so many different types of sensors on the market, it’s hard to choose the right ones. On top of that, relying on just one kind of sensor would be shortsighted because each has its strengths and flaws.
Intellias has written an overview of lidar and radar — two of the most popular sensors for autonomous driving technology. Let’s compare their main features and find out which works best with driverless vehicles.
Radar vs lidar for self-driving cars
According to McKinsey, so far there have been two main approaches to autonomous car perception: either a radar sensor combined with sonar and cameras or a lidar augmentation.
If you think of the map as having a static view of the world, the sensor system is providing a dynamic fill-in to that map. These two, together, provide what is called a ‘world model’ for that autonomous vehicle
In the autonomous car race, the lidar vs radar systems competition is getting real. It has even caused a kind of rivalry within the automotive industry. While many companies including Waymo, General Motors, and Ford rely mostly on lidar, Tesla has decided to stand with radar. Just recently, Elon Musk once again publicly shamed the lidar system technology for automotive, calling it “a crutch”.
So who’s right? Is lidar really unnecessary for autonomous cars? Can radar systems alone deliver accurate images of the car’s environment? Let’s dig into the main features of radar and lidar technologies and find out the truth.
Everything you need to know about lidar
Lidar is short for light detection and ranging. The technology uses near-infrared light to scan objects and create a 3D map of the environment. Here’s how it works. A lidar sensor transmits laser beams that bounce off objects and return to the sensor. Based on the information received, a lidar system creates a point cloud that looks like a shadow and reflects the object’s shape and size.
Lidar’s history starts way back in the 1960s. With the invention of the laser, laser-powered technologies became very popular. Perhaps the laser’s most famous and epic application was in the Apollo 15 mission, which used a laser altimeter to map the moon surface. Today, lidar technology has a lot of applications. It measures clouds in the atmosphere, maps the terrains of rainforests, and even guides SpaceX’s Dragon shuttle to the international space station. But most importantly, it brings autonomous car technology to life.
The global lidar market will reach $5.8 billion by 2024.
Global LiDAR market size in 2016 and 2024
Lidar system technology is by far the most accurate for autonomous driving
Lidar has a lot to offer to autonomous driving. It scans the car’s environment and creates accurate 3D images that cover a 360-degree view of the vehicle’s surroundings. A lidar sensor system can identify various objects. Using lidar, an autonomous car can not only see the road, vehicles, and pedestrians — it can also distinguish among vehicle types, cyclists, children, animals, and other objects that may require specific precautions like slowing down.
Lidar’s 3D images of the environment are so accurate that a vehicle can even see people’s gestures. Sounds impressive, doesn’t it? This will help autonomous cars get along with people on the road.
But most importantly, lidar can track movements and their direction. This is why the technology is vital for driverless vehicles. Thanks to lidar, a self-driving car will know who’s heading where. This makes it easier for a computer to predict the actions of road users and make decisions while driving.
Automotive lidar shipments will reach 34 million units worldwide in 2032.
Automotive LiDAR shipments worldwide between 2022 and 2032
The main challenges for lidar are price and weather conditions
Unfortunately, there are both pros and cons of the lidar technology. The first downside is that lidar sensors depend on weather conditions. They can’t deliver accurate images of surroundings in fog, snow, or dust. This actually means that lidar system technology always has to be paired with secondary sensors.
Beyond that, lidar isn’t good at detecting the speed of other vehicles. To make matters worse, a common spinning lidar that shoots lasers in different directions has a lot of moving parts. This makes it costly to both produce and maintain.
But probably the main disadvantage of lidar system technology is its price. Not so long ago, a single sensor made by Google cost $75,000. And even though the company has dropped the price by 90%, lidar sensors are still less affordable for auto manufacturers than any other kind of sensors.
This is how Google’s self-driving car uses a spinning lidar sensor to map the road.
Today, Google’s lidar costs $7,500 per unit. Following the company’s reduction in price, many other sensor manufacturers dropped the price of their lidar sensors too (take Velodyne, for example). The good news is that lidar’s price will only continue to decline. And thanks to solid-state technology, lidar units will become smaller and even more cost-effective.
In five years, for ride-sharing cars, it could be $8,000. In 2025, it could be $5,000.
What makes radar system technology popular?
Radar stands radio detection and ranging. The technology uses radio waves to measure the distance to objects as well as their velocity and angle. Similar to lidar, radar constantly sends signals that bounce off obstacles. It compares the transmitted signal to the received signal to understand the environment.
Radar technology was first invented in the nineteenth century but became popular in the early 1930s. Radar was first used to detect hostile aircraft. Today, radar technology is used beyond its initial military purpose.
Radar helps with planetary observation, air traffic control around airports, aircraft navigation, mine surveying, and even studying bird movements. And everyone knows the technology is popular for highway security: the police often use radar speed guns to track how fast a vehicle is moving. Moreover, radar is a trusted solution for autonomous driving.
Radar system technology is the most robust and reliable
Radar has many benefits for self-driving cars. It does a great job of preventing collisions, providing parking assistance, and powering cruise control systems. Unlike lidar, radar isn’t affected by weather conditions. Fog, snow, rain, and dust aren’t obstacles for its radio waves.
Also, a radar sensor is small, lightweight, and cheap. The cost of a sensor varies from $50 to $200. What’s more, a radar sensor doesn’t have any moving parts and requires less power than a lidar sensor. All this makes the technology very attractive for auto manufacturers.
Many automotive companies, including Volkswagen, BMW, and Mercedes-Benz use radar sensors in their autonomous cars. And Tesla went all in on radar technology for its semi-autonomous vehicles. Tesla believes that radar combined with passive optical sensors can do exactly the same job as lidar. Tesla’s radar system comes with eight cameras and twelve long-range ultrasonic sensors. But can a radar system fully replace a lidar sensor? Keep reading to find out.
Radar loses to lidar in terms of accuracy
Compared to lidar images, radar images have low accuracy and resolution. Radar can’t tell an object’s shape correctly. That’s why a radar sensor won’t allow an autonomous vehicle to distinguish all objects on the road. In other words, it won’t make any difference to a car whether there’s a deer or a tree ahead. For this reason, radar system technology is often paired with cameras and other sensor systems.
Sadly, there’s one more problem that automotive radar fails to solve. An autonomous car’s onboard ranging sensors can only detect objects that stand in their line of sight. If other vehicles are hiding behind big obstacles, the sensor won’t be able to see them and react quickly.
a. An autonomous car can’t spot the vehicle ahead because of its sensor’s limited range.
b. An autonomous car can’t see another vehicle when it’s turning since a wall is blocking the signal.
c. An autonomous car can’t spot a nearby vehicle because of the front-forward position of its sensor.
d. A radar system can’t see vehicles because another car is hiding them from the signal.
But there is a solution to this sensor system shortcoming. Autonomous cars equipped with vehicular communication systems won’t get into this kind of trouble. If a vehicle ahead shares its location, speed, and other valuable data, an autonomous car can fill in the gaps.
So far, the abilities of radar system technology are limited, meaning the information it delivers isn’t enough. Radar’s big advantage is that its data requires less energy for processing.
Yet the data from a radar system is more difficult to interpret than data from a computer-friendly lidar sensor. Make no mistake about it: driverless cars with higher levels of autonomy won’t rely entirely on radar. They need a more accurate solution to be able to see the world around them.
The good news is that radar sensors with higher resolution will appear soon. Right now, everyone is waiting for radar running at the 79 GHz frequency to hit the market. But short-range radar will still remain popular because of its reliability.
The global automotive radar market is expected to reach $7.9 billion in 2021.
Global automotive radar market size in 2015 and 2021
Does the difference between lidar and radar matter?
You bet! In essence, lidar can’t replace radar for autonomous driving, and vice versa. Both automotive sensor systems have their strengths and limitations. As for lidar vs radar accuracy, the answer is obvious.
A spinning lidar sensor is better at detecting objects. It provides good range and angular resolution. But its range is not as great as radar’s, which can see more than 200 meters. It also can’t measure radial velocity, which a radar sensor can. What’s more, radar is more solid and cheaper than lidar.
Comparison of environment perception sensors capabilities
In essence, there’s no need to continue the lidar vs radar fight for autonomous driving. These sensor systems can easily complement each other. And that’s just what many auto manufacturers are doing: they’re equipping their cars with both lidar and radar because where one sensor system fails, the other comes in handy.
For instance, in heavy rain, radar can back up a lidar sensor. And in a big city where an autonomous car has to see many details to drive safely, lidar technology will do the trick.
As you can see, there’s no winner in the lidar vs radar automotive battle. Both radar and lidar help an autonomous car map out its surroundings. But since these two sensors each have some flaws, it’s better to combine them.
By using data from various sensors, an autonomous car will be able to achieve higher levels of automation and safety. True, this will make the autonomous car more expensive, but when lives are at stake, it’s worth it.
The automotive experts at Intellias know exactly how autonomous cars can benefit from different sensor systems. Contact us to find out more about perception solutions for autonomous driving.