Omnitron Sensors’ MEMS could rid us of the spinning tops on self-driving cars (and lower costs too)

Omnitron Sensors, which makes MEMS sensor chips, has raised $13 million to create inexpensive sensors for self-driving cars. If it works, we could say goodbye to those big spinning domes atop autonomous vehicles.

The investment will fuel the expansion of Omnitron’s engineering and operations teams, accelerating the mass production of the company’s first product, a reliable, affordable microelectromechanical systems (MEMS) step-scanning mirror for multiple markets. MEMS have been used in everything from a Nintendo Wii to tire pressure sensors.

Corriente Advisors led the round with participation from longtime investor L’Attitude Ventures.

Streamlining the production of MEMS sensors — which have been limited by expensive, laborious manufacturing methods for decades — Omnitron’s fabrication intellectual property is the critical enabler for optical cross-connects (OXCs) in artificial intelligence (AI) data centers, optical subsystems in long-range LiDAR for autonomous navigation, see-through displays in extended reality (XR) headsets and eyewear, and precision laser spectrometry for methane gas detection.

The sensors can go into OXCs for AI data centers, and optical subsystems for advanced driver assistance systems (ADAS), drones, XR headsets, toxic gas-detection systems, and other connected electronics that are integral to our daily lives.

Sensor-driven markets

Omnitron is targeting major markets with its MEMS sensor technology:

• Improving throughput and energy efficiency in AI data centers: Omnitron’s MEMS-based photonics OXC for Tensor architectures boosts transmission speed and reliability in a low-power device, enhancing AI workflow in data centers. According to New Street Research, this market will approach $30B in 2027.
• Advancing affordability and reliability of LiDAR in autonomous vehicles: Omnitron’s MEMS step-scanning mirror targets an overall LiDAR subsystems market predicted to reach $6.3B by 2027, according to Yole Intelligence.
• Enhancing display quality in resource-constrained XR headsets/eyewear: Omnitron’s MEMS mirror satisfies growing demand for XR optics, a market that IDTechEx expects will exceed $5B by 2034.

“MEMS sensors are the intelligent microscale devices that allow us to touch the world through silicon,” said Aguilar. “Yet old and inefficient manufacturing methods have prevented the kind of growth in MEMS that istypical in semiconductors. That’s about to change. Our MEMS fabrication IP offers a new paradigm for mass-producing affordable, precise sensors at scale. With major investment from Corriente Capital and with additional funding from L’Attitude Ventures, our company can now deliver on the promise of our technology.”

Origins

Founded in 2019 by a core group of MEMS industry innovators, Omnitron Sensors has invented new MEMS fabrication IP that improves device performance and reliability, and that streamlines assembly to produce MEMS sensors for price-sensitive, high-volume markets.

Aguilar was pretty frank about how the company, which now has 12 people, got started. He used to work at companies like Google and Tesla, working on LiDAR systems. He thought of LiDAR as the “bane of my existence.”

He started the company with CTO Trent Huang, a quantum computing and MEMS manufacturing expert.

“I’m a nerd, and I’ve spent my career building or integrating sensors into robots so that they could see and operate in the real world. It’s working with these sensors, specifically LiDAR, that I saw the promise of what LIDAR could do for robots. But I also directly experienced the frustration with using LIDAR in real world applications. Even though billions of dollars have been spent in bringing LIDAR to the market, we don’t see it in every car. What is driving that issue is not necessarily a sexy word, but it’s reliability.”

He said the sensors often fail after just a few months of operating on the road. That has been the bottleneck for the big automakers, who don’t want to service a car for the sake of replacing a sensor.

Aguilar said, “That’s why LIDAR hasn’t made it into the market. And with my background in building kind of these technologies, and being a good engineer, I would visit these suppliers and go, Hey, what the heck’s going on?”

He saw a big game in the semiconductor fabrication processes and how to make them relliable. That’s why he formed the team five years ago and went to work on a plan to replace one of the moving parts with tiny little mirrors made with MEMS tech, where mechanical features are etched into semiconductors so that they have tiny little moving parts. In this case, they’re little mirrors on a chip that can move.

Now on its third round of capital, the company is refining its chips, which contain a 10-millimeter diameter mirror (that’s the gold part in the middle of the chips). It articulates 60 degrees, moving left and right, and it has a one millisecond settling time. That’s very fast, and it meets the requirements of a LiDAR system, Aguilar said.

“That has been what has really got our customers interested. And because of that interest, we were able to secure three letters of intent, two in the automotive industry, one in the energy sector, and valued at hundreds of millions of dollars,” he said.

How it works

LiDAR stands for Light Detection and Ranging. LiDAR systems measure distance by shooting a high-powered infrared laser at a target and closely measuring the pulse that bounces back. It works well whether it’s in the day or night.

A LiDAR sensor is a great solution in theory for a self-driving car. It senses the world around the car, which then analyzes whether there are any safety threats while a car is driving itself. It essentially reproduces the situational awareness that any human driver has.

But the LiDAR relies on a laser. It shoots out the laser light from the car. The light hits objects and bounces back, and that helps paint a picture of the environment in a digital form. The system receives the signal back and then calculates where the object is and in what direction it is moving. A processor takes all those points in a “point cloud” and then makes sense of all of the data.

Each LiDAR system has a laser. It also has a scanner, something that projects the laser into the world, and a receiver that gets the signal back.

The spinning device in a LiDAR sensor is scanning the environment, spraying out laser light and measuring the millions of points in a scene as the light bounces back. It’s scanning the environment. The mirror is akin to a big spinning disco ball. A photo detector is like a camera that captures the light.

It’s a moving component that spins the LiDAR around so it can get a 360-degree view of the environment. The challenge in a moving car is that this all has to happen fast because the environment changes as the car moves along.

The laser galvo, short for galvanometer, is an optical device that moves a mirror in response to electrical signals, usually in the laser galvo scanner. The movement of the mirror allows for precise control of the laser beam’s direction and focus. The MEMS chips can replace this device at a fraction of the cost.

“That’s the big disruption that we’re bringing to the market. And that’s what we’re changing inside of the the LiDAR unit,” said Aguilar said. “It’s a lot cheaper and it’s more reliable. That’s what silicon does.”

Progress toward production

To bring the product to market, the company had to design a new semiconductor process for manufacturing the MEMS devices to be smaller, cheaper and faster. The MEMS chip makers weren’t ready for this kind of advance, which is ten times more dense than others on the market.

“We had to go back to first principles and design a whole new technology node for MEMS to bring this to the market. And it’s a huge leap in in density. So we built a much denser chip that allows us to achieve this performance,” Aguilar said.

As an example, one feature dubbed the aspect ratio for the MEMS devices is normally about 20 to one, and for Omnitron Sensors, the aspect ratio is 100 to one, or a trench within the chip that is five times deeper than normal.

Now the company has been receiving its first chips back from low-volume manufacturing, and it’s testing the parts of reliability. The company is using foundries, or contract manufacturers that make chips in the U.S., to make the sensor chips.

But going from the “lab to the fab,” or from testing to fabrication, is where “a lot of these technologies and companies die,” Aguilar said.

Now the company is moving from building chips on a few wafers of silicon a week to thousands a month so it can prove the reliability for the automotive customers, Aguilar said. That means moving to bigger MEMS factories with higher capacity. That prompted the funding round.

For each vehicle, there are likely to be four MEMS mirrors, and two per LiDAR system. The MEMS sesnor is smaller, cheaper and doesn’t require as much power. There are rivals out there, including MEMS chip makers in China, but Aguilar believes he has a technological edge.

Besides cars, there are telecom, augmented reality and space communications devices that can use the tech.

Daily insights on business use cases with VB Daily

If you want to impress your boss, VB Daily has you covered. We give you the inside scoop on what companies are doing with generative AI, from regulatory shifts to practical deployments, so you can share insights for maximum ROI.

Read our Privacy Policy

Thanks for subscribing. Check out more VB newsletters here.

An error occured.