With ASIC-Based Approach, Altos Radar Races to Lead Automotive Vision
Launched in January 2023, Altos Radar is fast-tracking a viable 4D automotive radar system with a recent $3.5M seed round.
California-based startup Altos Radar recently announced its arrival to the advanced driver assistance systems (ADAS) market with the closure of a $3.5M seed funding round and the availability of its first product, a 4D imaging radar Altos V1.
The Altos V1, 4D imaging radar. Image used courtesy of Altos Radar
All About Circuits reached out to Li Niu, CEO and co-founder of Altos Radar, to find out more about this newcomer to the automotive market. The team of five began working on the product in December 2022 after leaving their prior jobs and formally started the company in January 2023.
Altos Radar Takes Advantage of Updated Frequency Standard
Prior to the 2017 opening of the 76–81 GHz frequency band, radars could not offer the needed resolution to act as a primary sensor in automotive designers. As such, developers relegated radar to velocity measurement or lower-risk applications such as rear-facing, cross-traffic alerts.
Altos Radar is capitalizing on the new frequency range standard by bringing its 4D imaging radar, operating at 77 GHz, to the ADAS scene. 4D radar provides the location of an object in three dimensions and adds velocity as the fourth dimension. The company claims its device offers image perception on par with the current crop of LiDAR systems. This makes the unit an economic option to replace or augment LiDAR and camera systems.
A Founder's Road to 4D Imaging Radar
With more than eight years of experience with autonomous systems, Li has had ADAS on his mind for quite a while now.
“Autonomous driving will be the single most critical, profitable market in the next few decades,” Li said. "One of the biggest roadblocks is the absence of high performance, low-cost, and reliable hardware for sensing and computing.”
In his work, Li saw an opportunity to take a novel approach to automotive vision: 4D imaging radar. Like many other founders, Li decided that a startup was the best way to realize his vision. Part of that vision led the team to an ASIC-based solution rather than a more traditional FPGA-based design.
Altos Opts for ASICs Instead of FPGAs
Most automotive radar systems today use FPGAs at their heart. The FPGA route offers plenty of advantages: the hardware can be upgraded later in the field, and development can proceed concurrently with manufacturing. However, developing complex FPGA code takes a lot of time, and high-performance chips tend to be expensive.
Altos Radar instead found an existing Texas Instruments ASIC originally designed for automotive L2 automation applications and optimized low-level code to get the most performance out of the chip. This approach sped up the chip's time to market and significantly lowered the unit cost beyond competitive offerings. While Altos Radar has not yet set the final pricing, Li said the company is expecting high-volume pricing to be in the $150 to $200 per unit range.
Prototype Altos V1 installed on a test vehicle. Image used courtesy of Altos Radar
The startup's solution—the only non-FPGA product in this category—handles computation on board to deliver real-time point clouds in a format similar to that of LiDAR systems. The radar offers long-range detection and a point cloud with up to 3,000 points per frame at 10 fps. The system can detect vehicles at 400 meters and pedestrians at 180 meters, discern objects that are as close as 0.31 meters from each other, and measure speed at 0.2 meters per second.
Camera, Radar, or LiDAR: Which Is Best?
Altos Radar claims the Altos V1 matches or exceeds typical LiDAR performance and has the advantage of greater environmental flexibility. Radar will penetrate fog, dust, and other vision obstructions that LiDAR and camera-based systems cannot. The Altos V1 also comes with a significant cost advantage, with advanced LiDAR units costing thousands of dollars each.
Compared to cameras, radar and LiDAR both require less computing power since they only stream useful information to the ADAS. Camera imaging requires significant processing to extract object identification, speed, and location before data is ready for ADAS.
The jury is still out on which of the three systems will rise to the top of the ADAS vision contest. There may be more than one winner. In the end, though, low cost, high performance, and time to market are critical routes to success—and in these areas, Altos Radar seems to have a solid start in the right direction.