In Part I of this series, we talked about Max Range and FOV. Closely related to FOV are Points Per Second and Frame Rate, which together determine a lidar’s actual image quality. Understanding the intricacies of Points Per Second is essential for assessing the suitability of lidar solutions for specific use cases and environments. Now let’s unravel the complexities of these specifications and their implications for lidar technology.
Points Per Second (PPS) and Frame Rate:
We know lidar images are point clouds, so Points Per Second refers to the number of points the lidar processes every second. It is a key indicator of a lidar's data rate, and some might think the two practically mean the same.
Theoretically, PPS can mean both the number of points that the lidar sends out, or the total number of points the lidar receives back from the objects and the points that never returns (those “lost” points are what maps out empty spaces). The higher PPS within a certain FOV, the denser the point clouds would look, thus the more details captured.
Therefore, PPSs are not directly comparable across lidars of different FOVs. Again, only PPS within the effective FOV is considered effective PPS. That means although a rotational lidar can have a high PPS, when it’s installed in a vehicle’s front grille or on the wall, with a significant part of its FOV hitting the ego-vehicle or the wall instead of the environment that needs scanning, its effective PPS is significantly smaller.
Another specification that needs to be looked at together with PPS is Frame Rate, which refers to how many times the lidar repeats the processing of an entire FOV within a second. Each frame is a lidar image, such as the black tire detection graphics in this blog post. When we process multiple frames, we get dynamic point clouds that allow us to measure the object’s motion.
With a given PPS and a fixed FOV, a higher frame rate results in fewer points in each frame of lidar image, but the images would refresh more frequently; a lower frame rate would allow each frame to have more points, but there would be fewer frames. Likewise, with a given PPS, to keep the same density of the points in each frame, a higher frame rate would be associated with a smaller FOV.
It is very much like cartoons – if the cartoonist is only allowed to put in certain hours of work, without changing the size of the drawing area, they would deliver either more slides of simple images or fewer slides of detailed images. Or they would create equally elaborate pictures but toggle between fewer larger slides and more smaller ones. In both cases, the quality of the cartoon would turn out differently in terms of choppiness, elaborateness or size.
Lidar company can find an optimal balance between PPS, Frame Rate and FOV depending on their lidar architecture and application. Some lidars need to detect object motion faster, some need to collect more details, and some need to “see” wider. Depending on specific use cases, different OEMs require different Frame Rates, but most commonly used Frame Rates are 10Hz and 20Hz, which are often configurable. Most lidar product marketing images are shown using data captured at 10Hz.
Angular Resolution and Range Accuracy:
Angular Resolution represents the average FOV allocated to each lidar point (or pixel if you will). Note here the key word is “average”, because depending on the scan pattern, some lidar images have points more evenly distributed than others. Some line scanning lidars with low numbers of channels may have significantly lower vertical angular resolution than the horizonal one, and that means their lidar images might suffer from large vertical gaps.
Our Vista-X lidar and Ultra lidar, for example, use different scan patterns. The former generates mesh-like point clouds, and the latter more linear-looking ones. While they both were invented to generate densely and evenly distributed point clouds, their angular resolutions are both average values, and randomly measuring the space between an individual set of points might result in slight deviations.
Range accuracy refers to how precisely the lidar can measure the true distance to a target. For example, a 2cm range accuracy means that the actual distance to a detected object might be 2cm more or less than what the lidar tells you. It’s kind of like very mild astigmatism, and naturally, different lidar use cases would have different standards for range accuracy. For automotive applications, the industry tolerates up to 2cm, while the flash lidar used to unlock an iPhone would require something much smaller.
Range accuracy mostly depends on the lidar’s detection method, component quality and overall data processing capabilities. Lidar manufacturers today choose between two common detection methods, Time of Flight (ToF) and Frequency Modulated Continuous Wave (FMCW). As FMCW is more susceptible to poor laser source stability (both optically and electronically) and more dependent on the quality of laser and detector modules, it enables significantly lower range accuracy per dollar than ToF. In other words, ToF offers a much more cost-effective solution while meeting OEM requirements.
In addition to the above-mentioned specifications that indicate lidar performance, size, power consumption and cost are also key things that determine the usability of a lidar solution. More importantly, those non-performance factors are direct indicators of how suitable and scalable the lidar technology is for mass-market applications.
Size:
A bulky size significantly limits the lidar’s embeddability and compatibility with modern vehicle design aesthetics. Not only does the lidar needs to be compact, but for many automotive use cases, the thinner, the better. High performance in a small package is enabled by an efficient lidar architecture as well as a powerful System on Chip (SoC), which optimizes signal processing for best results. Compared to merchant SoCs, proprietary lidar SoCs gets the job done in a more targeted manner, with significantly smaller footprint, power consumption and cost. That’s why Cepton has invested in developing our own lidar SoCs since 2019.
Power Consumption:
A powerful lidar solution should not be implemented at the expense of high power consumption. As OEMs are looking to reduce the environmental footprint of their vehicles, power-efficiency has become essential for lidar. Especially for electric vehicles, reducing the lidar’s power consumption helps maximize the vehicle's mileage between charges.
Low power consumption is also crucial for seamless vehicle integration, as the excessive heat resulting from high power consumption places constraints on the lidar’s placement options. Heat also leads to compromised sensor performance, premature wear and failure and safety risks – even an iPhone shuts down as protection from overheating, let alone lidar as a safety sensor in an enclosed environment.
Some lidar manufacturers integrate their lidar with an active cooling system – such as fans – to address overheating. Whether built into the lidar unit or not, the actively cooling system only dissipates the heat instead of eliminating it, not to mention the additional power and footprint it entails. At Cepton, we believe the best solution is to minimize the lidar’s power consumption to only require passive cooling.
Developing extremely power efficient lidars has been our focus since Day 1, and our Ultra lidar delivers superior performance at below 13W. That is equivalent to a 1500-Lumen LED light bulb or a small Bluetooth speaker. This makes Ultra seamlessly embeddable behind the windshield and in the roofline without the need for active cooling.
Cost:
Recent automotive trends indicate that consumers and car manufacturers continue to remain cost-conscious, especially during the EV transition. Needless to say, OEMs favor cost-effective lidars, but the extent to which they focus on cost reduction might be surprising.
Our core technologies minimize lidar cost without compromising performance and reliability, but equally important is our collaboration with world-leading, experienced manufacturing partners, which enables us to offer automotive volumes of top-quality lidar solutions at three-digit price points. These partners set an example of how much emphasis the automotive industry puts on quality – here’s a story told by our CEO Dr. Jun Pei.
To learn more about what we’ve previously discussed about automotive lidar, visit this blog post.
The complexities of lidar technology, as explored in this series, underscore the critical interplay between specifications like Maximum Detection Rage, Points Per Second, Frame Rate, Field of View, Angular Resolution and Range Accuracy. Each specification plays a crucial role in determining the efficacy of lidar systems across diverse applications, from automotive safety to consumer electronics.
While advancements in lidar architecture continue to refine these metrics, achieving optimal balance remains paramount for manufacturers. By addressing challenges such as size, power consumption, and cost alongside performance metrics, Cepton is committed to delivering usable, scalable lidar solutions poised to meet the rigorous mass-market demand.
