Self-driven cars are no longer a component of science-fiction stories. They have been under development for years and have already hit a few roads in the US under testing conditions. A large number of advanced technologies have gone into building the version of autonomous cars that we see today, of which, LiDAR is one of the most important ones.
LiDAR, which stands for Light Detection and Ranging, is a surveying technology that uses laser pulses to determine the distance of an object, often hidden behind obstructions. A LiDAR system typically consists of a light-emitting sensor that transmits hundreds of thousands of laser pulses in the surrounding.
These pulses, when reflected from the objects they hit, are registered by the sensor. Based on the time taken for the pulses to return, complex computing systems within the LiDAR unit decipher the distance of the objects that reflected the light.
Autonomous car manufacturers of this decade use this base technology to make such cars safe and reliable. Read on to know more about how LiDAR functions as the “eyes” of an autonomous vehicle, along with a story of where it all began.
A Brief History of LiDAR in Autonomous Vehicles
Back in 2005, the Grand DARPA (Defense Advanced Research Projects Agency) challenge saw the first usage of LiDAR is self-driven cars. The vehicle, named Stanley (created by Stanford University), deployed 5 LiDAR sensors along with cameras and other advanced technologies.
The sensors had to be precisely pointed in the right directions to capture the best shots of the road, and even then, they could only create a 2D view of the objects.
At the same time, Velodyne entered the foray and launched a 3D laser-based system for autonomous cars. By 2007, the company had perfected the functioning of the LiDAR system and also reduced its size. It was then being extensively used by teams that competed in DARPA.
The Modern Approach
The early inventions that teams explored in the DARPA challenge paved a way for the engineers today to build modern LiDAR systems for autonomous vehicles that are soon expected to overthrow manually driven cars. Here are a few approaches that today’s LiDAR system takes.
A LiDAR sensor is usually located atop a car, or in some cases on the bonnet. Either way, its job is to continuously rotate 360 degrees to send and receive numerous light signals that are then calibrated and used to detect the presence of objects around the vehicle.
These objects form the part of the 3D map that sophisticated computers within the car’s mechanism detect from the point clouds (data sets). Looking at the map, the computers can identify how far the object is from the vehicles and whether the object is moving or stationary.
The algorithms can also calculate the possibility of collision with a moving object, for example, another vehicle or a pedestrian. The features of LiDAR enable the car to scan objects at a higher resolution and long ranges. Therefore, it is easy to prevent collisions and enjoy a safer ride in a self-driven car.
Knowing what lies on the road that you are driving on does not need guesswork anymore. With LiDAR, pre-scan modes can draw a 3D map of the road ahead, for a long distance.
The pre-scan system allows the light sensors to send and receive multiple pulses in every direction of the road. By doing so, the car gets to know all the obstacles it needs to avoid, such as the narrowing or widening of roads, signs & posts, lane boundaries, traffic lights, under construction areas, road repair work, potholes, trees, and much more.
The best part is that LiDAR can see through fog and mist so you do not have to worry about visibility issues when traveling in a self-driven car under unfavorable weather conditions. This method of predicting the road makes it less dangerous to travel on unchartered territories and is a boost for adventurers and explorers.
Advanced Driver Assistance Systems (ADAS)
The most advanced approach to build a completely safe autonomous vehicle is to deploy advanced driver assistance systems. An ADAS is typically designed using a combination of vision, RADAR, and LiDAR sensors.
Vision sensors are used for high-visibility situations, such as for parking assistance, identifying traffic signs, reading stop signs, recognizing markers on the road, and so on. In low visibility conditions, RADAR-based sensors are effective is providing visual assistance to the ADAS as they can cover longer ranges. When it comes to mapping the surroundings and generating a 360-degree view outside the vehicle, the ADAS relies on the LiDAR-based sensors.
Complex machine learning and image processing algorithms process the data collected from all these sensors. Thus, the three systems work in tandem to provide complete navigational assistance to the automobile, providing total control over the steering, the brakes, and the speed.
Types of LiDAR Sensors
LiDAR sensors used in modern automobiles are of two types. The types differ in the technology used and may differ in positioning as well. Here is more information about the types.
– Electro-Mechanical LiDAR
This is the kind of LiDAR typically seen on top of vehicles. These are made from multiple moving parts and weigh a lot. They are the ones that rotate at all angles and throw out pulses of laser in every direction.
Such a kind of LiDAR system is expensive and hence and a lot of automobile manufacturers are reluctant to use these systems until the cost comes down. They can also be difficult to repair when they undergo wear and tear because of multiple mechanical parts.
– Solid-State LiDAR
Solid States LiDAR systems are more compact and are the future of LiDAR systems used in self-driven cars. In such systems, all the parts, including the transmitter, the receiver, and the processor are built on a single microchip.
This chip, being much smaller in size than an electro-mechanical LiDAR system, can be installed anywhere on the vehicle without taking up the bulk of the space up top. These are easy to maintain and are less prone to wear and tear as there are no mechanical parts involved.
As the world of automobiles gazes into the not-so-distant future of autonomous vehicles, the focus is on lowering the cost and improving the resolution and the range of the 3D maps. Solid-state LiDARs provide some answers to the big cost question, but they also come with a disadvantage.
If you are using small, fixed sensors instead of large, rotating sensors, you may have to embed multiple sensors into the cars, and even then, the range of the sensors might not be good enough. Hence, manufacturers are looking at efficiently optimizing the use of both types of LiDAR sensors by combining their advantages.
High-resolution mapping is an area that engineers are still working on and is expected to be the focus of automobile innovation in the coming years. With all these advances in place, safe and economical road trips on self-driven cars are slowly inching towards perfection and are expected to be a ground reality in a few years.