Featured Founder: Spartan Co-founder & CEO Nathan Mintz

This month, we feature Nathan Mintz, co-founder and CEO of Spartan Radar.

What made you decide now was the time to pursue autonomous vehicles? What inspired you to start Spartan?

My interest in automation is rooted in a career working on systems designed to protect the American warfighter. I split my career between building sensors to help our warfighters and building electronic warfare systems to defeat the adversary sensors in kind. Some of these included the VIIRS weather sensor that tracks weather and wildfires from space, the APG-79 AESA radar on the F-18 fighter jet, various sensors for the Global Hawk and other UAVs and the Next Generation Jammer (now ALQ-234) program.

In mid-2020, I was starting a venture studio with some other entrepreneurs (Dangerous Ventures, which also helped incubate Spartan). While there, my long-time mentor and Spartan co-founder Dr. Theagenis Abatzoglou approached me with an idea he had for a new type of automotive radar based on some novel areas of signal processing he had made a career out of: super resolution and compressive sensing. This technology would result in 5x greater resolution while using 1/4 of the transmit/receive elements (a key cost and complexity driver in every radar).

Believing we now had some game-changing technology, I now wanted to take it a step further by going deep, then looking at it from a systems perspective where you “fall in love with the problem, not your solution to the problem.” I dove into the problem to understand the autonomous vehicle industry by learning from industry veterans, like our co-founder and CRO Blake Gasca, and I came to two key realizations:

• The first was that the industry was heavily rooted in (and therefore biased from) its background in “big data” machine learning.
• The second was that they were still treating sensors the same way aerospace did in the 1960s: as single function cameras with little flexibility to change their focus based upon environmental conditions (what we called use cases in aerospace, operational design domains or ODDs in autonomous vehicles) and minimal ability to weight data for context. This dynamic resource management was the key realization behind “biomimetic radar.” Part of the reason for bringing co-founder Tyler Rather in as CTO was his expansive background in these kinds of resource management, AI, and signal processing algorithms he’d developed in the 15+ years we’d worked off and on together.

Pitch me: Why is radar a better choice than LiDAR?

Well, Elon Musk famously called LiDAR a “fool’s errand” but I do think that for some domains it has a lot of utility. If I want to map the interior of a building in three dimensions with a drone quickly or dock with someone in orbit with millimeter precision, a LiDAR may make good sense. But today, fully automated vehicles are over reliant on LiDAR. The industry now realizes to scale from hundreds of R&D vehicles to thousands of production vehicles, they must integrate smart radars that have performance metrics similar to LiDAR with the robustness of automotive radar.

The reason is LiDAR isn’t robust — it has major problems with weather, smoke, and even condensed exhaust that cause a lot of unpredictable gaps in performance. Range can fall off 50% or more in rain or snow for LiDAR. Combine this with its high cost (thousands or more per unit) and reliability issues, and LiDAR is even more problematic. One example is that the commercial trucking applications are replacing their LiDAR every 10,000 miles; this is NOT commercially scalable. Automotive grade radar on the other hand is low cost, works in all-weather conditions, is relatively low power and is extremely reliable.

Developers keep falling back on LiDAR, because it has excellent angular resolution. However, when you ask them “Well, how much resolution is good enough to deliver people and goods reliably and safely?” In conjunction with cameras, they repeatedly answer “about one degree in angular resolution is where I could think about replacing the LiDAR with the radar.”

This is the pain point Spartan addresses: most automotive grade radars coming to market today are 2 degrees (or more) in angular resolution, not good enough to abandon the LiDAR. Our software upgrades those same radars so they can get sub 1 degree resolution, all while running on the edge in existing processors. We also want to enable this to happen faster and more dynamically to maximize processing resources and relevance of information.

We make everyone’s radar better with our software products called Ago. We want to be the intelligence behind everyone’s radars that takes them from single function sensors into dynamic sensors that work complimentarily with the whole system to make it safer, faster reacting, more processing efficient and more reliable.

Have you always been a car person?

To be honest, I drive a nine-year-old Camry. However, what gets me excited about the automation of vehicles is I am the father of four young children who I want to be able to drive safely everywhere they need to go in the future.

Along with the added safety, I am also incredibly interested in sensors and the autonomous systems that keep them running and how we can improve them to exponentially improve our productivity and quality of life. I have spent my whole career inventing new things to do with RF sensors and effectors, including the last company I founded Epirus (also a PML investment) where we used high power RF to defeat drones and other electronics.

We are entering a new era where autonomous sensors and effectors will become ubiquitous and weave themselves seamlessly into the fabric of our lives. I heard a presentation from the CTO of NVIDIA a while ago where he predicted we would exceed 1 trillion IOT devices by 2030 — that’s over 100 autonomous devices for every person on earth. The implications are mind bending.

Why is this the right time for autonomous vehicles?

I think it is the right time for ADAS systems — that is what the autonomous folks call SAE Level 1–2 or driver assistance. This is where the platform aids the user to enhance safety, improve productivity and fuel efficiency. Examples of this include Tesla’s FSD, GM Super Cruise, and many of the safety systems being introduced in trucking and mining fleets. In these solutions you have “human in the loop” and these are becoming ubiquitous in most all vehicles today.

Level 4 autonomy, where the human is basically designed out of the loop under a certain ODD, is more like 3–5 years away, in my opinion. The big issue here is the need to scale L4 vehicles from R&D to production. I believe, along with many people in the industry, that this will occur in the movement of goods first. Whether that is an Aurora autonomous truck delivering for FedEx or a Nuro delivery vehicle dropping off a Domino’s pizza in your neighborhood.

Over the past year you have seen the larger AV companies raise billions of dollars in the private and public markets. The AV industry is no longer dominated by small startups, it is well capitalized and most are partnered with traditional OEMs to deliver thousands of autonomous vehicles by mid to late decade.

At Spartan, we have product offerings for all the levels of autonomy. We drive to be the intelligence behind all the automotive radars regardless of if it is an L2 vehicle you drive off the lot next year or a UPS truck powered by TuSimple driving cross country driverless in the next several years.

What was the least expected challenge you’ve overcome to reach this point?

We had a co-founder quit four months in to take a research job at GTRI. He gave me two weeks’ notice out of the blue. He said he realized two things: 1) he wasn’t an application guy; he was a research guy; 2) he wanted out of California. We parted amicably but it still required some adjustments to not have it break the momentum. However, with his departure we have been able to recruit top leadership talent out the AV industry, aftermarket, and automotive.

Who inspires you?

My wife inspires me. She manages to successfully juggle four kids under ten while putting up with her husband’s crazy entrepreneurial lifestyle, and she does it magnificently. She’s a former Raytheon radar systems engineer herself who is very savvy, and if I had a dime for every time she was right about something I was struggling with and couldn’t quite find the solution, I could skip the Series B.

What social causes do you care about and why?

I was involved over a decade ago heavily with California politics, including two runs for assembly in my mid-20s. The desire to make government more transparent and accessible to all citizens led me to co-found a non-profit with Joe Lonsdale called California Common Sense back in 2010 (Dakin was its first executive director). The tools developed for CACS eventually turned into start up OpenGov.

I’m incredibly passionate on causes that drive power to the constituents not the politicians. I believe our elected leaders need to serve the people that put them in office, not the special interests or well connected. That is why I’m an advocate for redistricting reform, and I actually served on a redistricting commission for the local community college district where I live a while ago.

Self-driving cars were pretty hyped a decade ago. What’s a more realistic 10-year plan look like?

The first domains to commercialize will be in mobility of goods, like trucking, mining, drayage, warehouses, and others because they have the most acute pain point. ATA estimates we have an over 80,000 truck driver shortage in this country right now and that will only become more acute as baby boomers move into retirement, and as a society, we have shifted consumer habits to an on-demand delivery economy. People don’t realize that trucking in the US is a trillion-dollar industry and growing, over 70% of all goods were moved by a truck this year.

In terms of the car in your driveway, I think we are going to see things become more and more autonomous step by step, first enhancing driver assistance to make them safer, then growing into greater automation in certain use cases (e.g., trucking, last mile delivery) limited domains and then finally culminating in fully autonomous cars deployed in ride hailing service model by the end of the decade.

What are the inflection points we should be looking for in the industry?

I fall back to the 5% adoption rule: once a product becomes adopted by 5% of users in a particular domain, it starts to grow exponentially with 25% and 50% following even faster. We saw this with the internet, cell phones, and other commercial goods. Electric vehicles in general are also nearing that inflection point.

We are a couple years away from that for imaging radars (the first ones are just starting to be adopted on a few high-end cars like the Mercedes S-class and some commercial trucks) but our Spartan Ago software can make that happen faster by reducing the complexity and allowing us to get there with commercially available processors available today. Once 5% of the vehicles in the road have imaging radars, I can see them becoming ubiquitous very quickly thereafter and that will happen within the next 5 years.

Have you read anything lately that inspired you?

I recently read “Play Bigger” by Christopher Lochhead & co. (you may have heard the “Lochhead on marketing” weekly podcast) at the recommendation of one of our marketing consultants. The book lays out a roadmap for category design — how a company comes to define the market on their terms and how to create a “product flywheel” drawing on examples like IBM in the 50/60s, Microsoft, Apple, Google, and Facebook. As a recovering engineer, my intuition was to dismiss all this meta marketing about category creation as michigas and hocus pocus, but over time I found it to be an integral part of creating a great business and have endeavored to integrate it more into my thinking.

You’ve recruited a great team. What are your secrets?

I have been lucky to have worked on a lot of different things and to have met a lot of brilliant and talented people doing so. My wife used to joke that it took me two hours to get to my desk when I first started at Raytheon because I had to schmooze with four different people in the hallway on the way there. I have a rule to keep in touch with people because you never know when your paths will cross or the stars will align and you need them. I guess you could say I’m a people person and my network has been built on trust and success.

As Jim Collins’ put it “get the right people on the bus…then find them the right seat” — the first is telling them the story that your idea is exciting and will be a winner — that their stop is at the end of the route. “Finding them the right seat” is a little harder. People often make mistakes finding the right candidate but putting them in the completely wrong role — like taking an amazing programmer and making them a manager. I’ve made this mistake myself a lot of times, but as a result, I put a lot of thought into crafting a role where this particular person will be uniquely suited for.

The last piece is to be generous and think big. This is a lesson the chairman of my last startup’s board actually beat into my head. “If you think this candidate could unlock millions in of value, why are you risking blowing up their hire over a small bit of salary?” he once told me. Early stage is all about growth and if you don’t put the right team on the field early in the season and start winning quick, it won’t matter that you bought yourself an extra week of burn by driving the hardest offer you could by over-optimizing to get the absolute best value possible.

Prime Movers Lab invests in breakthrough scientific startups founded by Prime Movers, the inventors who transform billions of lives. We invest in companies reinventing energy, transportation, infrastructure, manufacturing, human augmentation and agriculture.

Sign up here if you are not already subscribed to our blog.