The Sun Bay Paper Page 18 March 18, 2022 - March 24, 2022 Self Driving Vehicles: RADAR vs LIDAR In 2020, a report by Statista predicted that 1 in 10 vehicles on the world’s roads is expected to be autonomous (self-driving vehicles) by 2030 with no distinction of electric or fossil fuel power. According to the report, that means manufacturing more than 100 million autonomous vehicles in the next nine years — and growing sales volume to almost $14 billion. Today, with roughly 40 states having enacted legislation or executive orders allowing for automated delivery services to operate, those numbers start to look conservative. The stats don’t lie — the unprecedented rise of the autonomous vehicle is just beginning. In fact, the more you look the numbers just keeps getting larger. Analysts believe that the global autonomous car market will swell to nearly $32 billion in 2022 and then surge to $37.2 billion in 2023. Between their transportation and delivery applications, autonomous vehicles could potentially capture a large share of our $14 trillion consumer spending economy, But would you trust one enough to go for a ride in one today? THE ROAD TO SAFER AUTONOMOUS VEHICLES Securing public support is a pivotal catalyst for the adoption of autonomous technology.And while 40% of people worldwide are willing to use a self-driving vehicle, it’s clear that there is still some hesitation when it comes to safety, we can hardly blame them for that, especially considering some of the headlines that have made their way to the front pages over the last few years. Uber’s prototype self-driving car was involved in more than 35 crashes between 2016 and 2018 — one of which killed a pedestrian in Arizona. And to be clear, that crash — along with a few others — was the fault of the vehicle and its backup driver. Autonomous vehicle companies have since implemented a number of technological and operational changes to prevent such tragedies in the future. But the Department of Transportation’s data reveals an interesting thing about autonomous vehicle collisions — the overwhelming majority are actually the fault of other drivers. And the number of accidents appears to be steadily decreasing over time. From 2018–2020, Alphabet’s Waymo self-driving taxi service, now serving riders in Metro Phoenix, AZ. (formerly the Google self-driving car project) was involved in 18 collisions. That’s just half of Uber’s accident count from the preceding two years — and Waymo is a fully operational selfdriving taxi service available to consumers, while Uber was just testing prototypes. The implication is clear — autonomous vehicles aren’t so unsafe after all. They’re getting better at driving and thus safer over time. That’s largely thanks to advances in LIDAR in the last few years. THE SECRET BEHIND LIGHT DETECTION AND RANGING (LIDAR) LIDAR, as its name implies, is a very similar technology to radar. In fact, the only difference is the wavelength of light being used to identify surrounding objects. Radar on one hand, uses radio waves, which have very long wavelengths that often measure several meters. This means they can travel very long distances, making them useful for military purposes like detecting aircraft or boats in a large area. Because they generate a relatively low volume of data points, radar systems don’t require much in the way of advanced hardware or software to be useful and can thus be produced and installed quite cheaply. Many automakers have integrated radar-based cruise control or collision-avoidance systems into their vehicles. In fact, more than 92% of new vehicle models on sale in the U.S. in 2018 had some such radar-based system, according to AAA. The proliferation of these “proto-autonomous vehicle” systems has been credited with a significant reduction in U.S. road deaths in years past, despite a significant increase in the country’s population and average driving miles per person, but radar has its flaws — especially at the small scales relevant to urban driving. The long wavelength means that the “images” returned by radar sensors are extremely low resolution. Most radar displays simply mark the location of objects with dots or patches, giving no visual detail and radar cannot reliably detect objects that are smaller than its wavelength — so while it’s able to spot large vehicles on a highway, it’s not quite up to the task of spotting children or animals crossing a busy street LIDAR solves this problem by using higher-frequency light waves — which are just a few nanometers long — to generate tens of thousands of detailed data points about the surrounding environment. It then uses algorithms to assemble those data points into detailed, highresolution maps. Compare the image above, which depicts what a marine radar sensor sees, with the image below, which depicts what a carmounted LiDAR sensor sees… This level of visual detail has helped Waymo cut collisions in half, while still scaling up the number of trips it makes, and its applications aren’t limited to experimental autonomous vehicles. For instance, The Waymo Taxi’s and there’s already a market for LIDAR drones for military use. The technology can be a game-changer on the battlefield. LIDAR allows troops to be better prepared and take advantage of highresolution maps of the terrain and know the exact position of enemy forces. Unfortunately, the costs of LIDAR is high, and as we previously mentioned, LiDAR is much more computationally intensive to use than radar. In addition to the sensor apparatus itself, a reliable LiDAR system needs robust hardware and software that can process the sensor data into a map — and do it fast enough to respond to sudden changes around the vehicle. Plus, the lasers LIDAR systems use to explore their environments aren’t exactly cheap. The industry-standard 1,550nanometer laser requires exotic materials like indium phosphide, which are only available through nonstandard, custom supply chains. As a result of these challenges, LiDAR systems have usually carried a fivefigure price tag, with $75,000 per sensor often cited as an average cost. But today, one obscure tech company has figured out how to produce military-grade LiDAR… for 99.8% cheaper and given its growing dominance in the trucking, drone, and autonomous-vehicle sensor industries, it seems destined to be the company that makes LiDAR into a household name. VELODYNE LIDAR (NASDAQ: VLDR): MAKING ROADS of the FUTURE SAFER Velodyne Lidar is one of the earliest pioneers in LIDAR technology. Founded in 1983 and based in San Jose, California, it developed the first 360-degree LIDAR sensor back in 2007, when self-driving cars were almost entirely reserved for science-fiction movies. Cont. pg 27
RkJQdWJsaXNoZXIy MjA2ODE3