Lidar and Autonomous Cars
Introduction
An autonomous car is one of the most complex technologies that an ordinary citizen can own. It is a powerful combination of electronics, innovative software, networking, and mechanical parts. Intricate electronic sensors go hand in hand with complex software algorithms to control its autonomous car and its functioning and communication with other nearby autonomous vehicles. The Society of Automotive Engineers (SAE) has defined five levels of driving automation. According to it, automation increases with a decrease in human engagement with the vehicle.
Lumenci had previously published a blog on Autonomous Cars: The Future of Mobility, where we discussed various functions of an AV. The blog discussed how a complex system involving processors driven by software, LIDAR, Radar, safety systems, Ultrasound, Video systems, and much more works together to ensure an AV's safe and successful functioning.
In this blog, we dive deeper into one of the most critical sensors used in most Autonomous vehicles - Lidar. It is a crucial sensor as it acts as the eyes for most autonomous vehicles today. Let's see in detail how Lidar works and explore companies that have built strong IPMoats around this technology.
Lidar
Light Detection and Ranging (Lidar) has become the most popular active sensory with the advancement in self-driving technology. Lidar is used to generate a high-resolution 3-D image of target objects. As the name suggests, it uses a laser to determine the distance between the target objects and the Lidar. It is a non-contact range-finding technique in which a short optical pulse is guided onto an item, or target and the backscattered pulse is spotted and processed to determine the distance to the target. Lidar does these millions of times per second. This data is further processed to develop a 360-degree image of the environment. Like humans rely on their eyes for vision, most self-driving cars rely on Lidar. Lidar builds a detailed 3-D representation of the surrounding by measuring the distance between the autonomous vehicle and the surrounding objects such as other vehicles, trees, people, roadblocks, etc. Lidar ensures safe and efficient navigation of an AV by preventing clashes between the car and the impediments.
Application of Lidar
Lidar is getting popular with Autonomous vehicles, but Lidar is not a recent technology. It was invented in the early 1960s after the invention of the Laser. It was even used in the Apollo 15 mission in 1971 to map the surface of the Moon. Nowadays, we can see Lidar technology on mobile devices as well. Apple's iPhone 12 Pro comes with inbuilt Lidar for better depth sensing. Lidar is also used in the agriculture sector for functions such as crop yield analysis and field management. Lidar can be used to provide 3D models of the terrain with the help of high-resolution data sets. Many companies use this function to perform mapping of fields covered by the vegetation in forest areas.
Working of Lidar
For imaging, Lidar uses Time-of-flight (TOF) principle. According to this principle, the distance between the target and the light source is measured by calculating the delay between the light emitted and light reached back to the source after reflection. It helps in forming a 3D point cloud of the environment. It can also be accomplished by modifying the signal's power, frequency, or phase and calculating the time required for that signal to arrive back at the receiver end.
To determine the distance between the target and the Lidar, a pulsed approach is used. An elementary measurement approach involves emitting a short light pulse towards the target and measuring the time taken by the pulse's echo to reach the detector. Let us understand this pulsed approach with the help of a block diagram.
In this technique, a small portion of the pulse is directed toward the Receptor start, which will start the timer. The rest of the pulse is transmitted in the direction of the target. The timer stops counting when the same pulse reaches back to the Receptor Stop after reflecting from the target. The time calculated is further multiplied by the speed of light in a (given) medium. Since the speed of light is constant in the same medium (air in this case), the distance to the target is directly proportional to the time traveled. This traveled time is double as light travels to the target back and forth. Therefore, to get the actual distance to the target, the speed of light must be divided by 2.
D = T/2*c
D is the distance to the target,
c is the speed of light (c = 3 × 10^8 m/s) in free space
T is the pulse's time to reach the target from its transmitter and then back to the receiver.
Maximum distance D that a Pulsed Lidar sensor can measure depends on the following factors:
Size of the counter used to record the time T
The size of the counter is at a maximum value that can be reached or counted by the counter before getting reset to zero.
The intensity of the laser pulse
When a light pulse is emitted and directed toward the target, only a portion of the pulse transmitted can reach back to the Lidar. After reflection, the amount of light that goes back to the Lidar depends upon three main factors: - the transmitted light intensity, how far the target is, and the size of the target. For the sensor to work, the pulse received back at the detector of the Lidar should be more than the minimum pulse intensity requirements of the sensor. Thus, the power of the transmitted pulse decides the maximum distance the Lidar can measure. To increase the distance the pulse intensity needs to be increased proportionally, and however, it can only be increased up to a certain level. The maximum light intensity is capped by the average power that is safe for the human eye as high-intensity light is harmful to the human eyes.
The pulsed approach is simple in setup, but it has a low signal-to-noise ratio. Other than the pulsed approach, Lidar can also measure frequency or phase shift to speed or distance.
The data (i.e., distance) collected is represented as a pointwise measurement to create a 3D image. Technically the measures (points) of the LIDAR are gathered to create virtual images of the impediments. The vehicle's autonomous driving system uses the collected data and tracks it in real-time to enhance its awareness of the surroundings.
The difference in Lidar, Radar, and Camera
Lidar provides the complete 3D perception of the surroundings instead of the conventional 2D projection from a camera. However, the performance of Lidar depends a lot on environmental factors. For example, the accuracy of the Lidar degrades in the foggy or rainy season. For these situations, RADAR (Radio Detection and Ranging) is an excellent solution. Its functioning is very similar to the Lidar, but RADAR uses Radio Waves instead of Laser to measure the distance. RADAR cannot develop a detailed image of the surroundings around the autonomous vehicles as efficiently as Lidar, but it can work in different weather conditions. Radar is also excellent in measuring the target's velocity and is therefore used in identifying the momentum of other cars or vehicles on the road. This dramatically helps an AV in controlling its speed concerning other cars.
Though Lidar and RADAR sensors provide critical functionalities of an AV, such as detailed imaging of the surroundings and identification of the speed of other vehicles, and AV also requires cameras. Cameras are necessary components for an AV. They enable it to detect traffic lights, other vehicles' indicators (left/right), traffic signs (speed limit), and many other things on the road. Not only this, but a complex framework involving multiple cameras can also perform specific tasks that Lidar conventionally performs.
Tesla v/s Waymo over Lidar
Tesla and Waymo are two key players in the autonomous cars industry, but their approach towards using the Lidar is entirely different. On the one hand, Waymo and other major players (like Aurora, Ford, Lyft, etc.) use multiple Lidar sensors to detect everything around an autonomous car. Still, on the other hand, Tesla's autonomous cars do not have even a single Lidar. Tesla uses multiple cameras supported by complex neural networks for visual recognition and detection of the surroundings. At present, both companies have lots of research and development, and at this considerably early stage, it is hard to decide which technology will emerge as the leader. However, in a matter of time, we can experience both technologies and decide which is the best.
Promising Start-ups
Several start-ups are now emerging in the field of Lidar. Their invention and technological advancement in this technology domain seem very promising:
Lumotive: It is a Seattle-based start-up founded back in 2018. It is using a patented Liquid Crystal Metasurfaces™ technology for its solid-state Lidar-based solutions. They offer different products based on Lidar for automotive, smartphone, and industrial applications.
SOS Labs: It is another start-up that delivers solid-state scanning in their Lidar. This South Korea-based start-up uses its invented MEMS + Hybrid architecture for 3D scanning. They provide three products: - ML (Mobility Lidar) used in autonomous vehicles, SL 3D Lidar for the Surveillance, and GL 2D Lidar primarily used for industrial applications.
SiLC Technologies: It is a California-based start-up that provides a fully integrated single-chip Frequency-modulated continuous-wave (FMCW) Lidar solution. Like the Pulsed approach (discussed earlier), FMCW is another approach used for Lidar imaging.
IP Moat
Lidar technology is witnessing rapid growth in the patent filing trends across the globe. Over the past few years, the number of patents published doubled year by year.
Patent Publishing Trends over 10 years (No. of patents)
(Source: Lumenci)
Bosch is a crucial player in the market. So far, it has filed more than 426 patents. Ford Global Tech LLC comes second with 262 patents filed.. General Motors comes fifth with 157 patents. These companies have built strong IPMoats in US jurisdiction.
Other key players who have filed for patents in Lidar technology are LG electronics inc., Honda motor co. Ltd., Alphabet Inc. (Google's parent company).
Top 7 Patent Assignees with number of patents for Lidar
(Source - Lumenci)
Conclusion
Lidar technology plays a critical role in Autonomous Vehicles and shows its use in different areas like land surveying, photography, etc. But its high price is a crucial factor that comes in the way of Lidar's use in day-to-day life. However, R&D in this field is currently going on with key objectives to reduce the price and increase the efficiency of Lidar. As we are moving near the Level 5 of the Autonomous vehicles, we are positive to witness rapid growth in Lidar development.
Author
Sourabh Thappar
Associate at Lumenci
Sourabh is an Associate at Lumenci. He holds a master’s degree in VLSI Design from IIT Dhanbad. He is interested in Portfolio Analysis and Damage estimation and is inclined towards Semiconductor and Wireless Communication. He likes to read about new technologies. He enjoys watching cricket and listing to music for leisure.