Silicon Valley venture capitalists like to talk about “filling the funnel.” What they’re saying is they have to look at upwards of 500 companies every year just to invest in 10. There are no short cuts. No matter how stringent your screening process is, you can’t cut it off at 100 or 200 when you’re looking for a market-disrupting technology. The math behind statistics means that there is no substitute for "sample size" if you want to see the shape of the bell curve, and identify the outliers at the top.
So you’ve got to know Silicon Valley, you got to know the top players, the upstart companies, where the potential resides. It's no longer enough to simply build a fantastically engineered vehicle. OEMs must have the software and connectivity services built in if they want to compete. So, they want to be first to find these ideas.
And as emerging technologies have become strategically vital to the ongoing progress of the automotive industry, that’s what the AutoTech Council is there for. “We built this organization to help them understand the Valley, follow best practices among other carmakers and Tier 1 vendors that have set up offices in the Valley,” said founder Derek Kerton. “But one of the most importantly things we do for them is find startups, and help them fill the funnel. We help them manage the Valley.”
Kerton spoke with leaders from four prominent mobility startups to get a sense of how they are managing the Valley.
Maha Achour — Metawave Corporation
"In 2017, Metawave’s radar technology was spun out from Xerox PARC. We are building the long-range radar with a 4G rate and imaging capabilities. At the same time, the solution that we are developing addresses 5G and wireless communications issues. Our WARLORD radar detects automobiles and their speed at greater than 300 meters and pedestrians beyond 200 meters. It's important to go beyond that distance because if you want to avoid casualties or fatalities, the car has to see. It has to perceive the environment at this long distance.
“Ours is the only sensor that can detect the car or pedestrian and also measure the velocity. We also have an AI engine, so we distinguish between cyclists, pedestrians and cars, even in low-sight conditions like fog. The WARLORD platform can also look around corners and detect a motorcycle without it being in the line of sight."
"We also will have the capability in the future to learn to proactively create productive analytics to detect a road issue and communicate vehicle-to-vehicle to another WARLORD radar.”
Bert Fransis — Arbe Robotics
Today, autonomous driving based on camera-only systems is just not feasible. The camera has advantages, but a lot of disadvantages as well. For example, it cannot measure speed very accurately, it cannot see in fog. A sensor is needed that can effectively see day and night, penetrate dust, rain and fog at a long-range, and make occluded objects visible. Most importantly, it has to measure velocity directly and in real time. And it needs to be affordable—LiDAR is too expensive for the mass market.
“For a truly autonomous driving solution with sensors, this is the radar checklist and the radar that Arbe Robotics has developed. We have a very wide field of view in real time with very high frame rate, so it can automatically be merged with a camera. It is highly accurate in terms of range and velocity, as well as in with elevation resolution. Our solution can detect out to 300 meters and, very importantly, it minimizes false alarms and false negatives. It also tracks all the stationary objects, which today's radars cannot do, and can see out of line, out of sight objects.”
Emmanuele Spera — NEXT Future Transportation Inc.
"We are an AI and robotic startup commercializing an electric, autonomous, modular vehicle and the operating system needed to transfer the transportation-as-a-service that we are providing to markets at scale. We are looking to solve the congestion problem. It doesn't matter if you have a traditional car and ride-sharing or a self-driving car. You'll still have congestion because you're using cars. Our solution is a mix of the ride-sharing experience, a mix of a car and a bus. We call them E-pods. They accommodate 10 passengers and are modular, coupling, uncoupling and being hailed via an app on demand—six pods connected equal a traditional bus.
“We've been testing pods since the beginning of 2018 in Dubai. We are on track with a prototype. The next generation of prototype which will be safety-validated with European standards. The technology's growing, building our pattern portfolio."
"Our projection is that we’ll be operating in Dubai by 2020. We’ve had interest from transit authorities around the world, and we're also working on logistics to use the pods as a last-mile delivery solution.”
Chang Yuan — Foresight AI Inc.
“We're building a global-scale data platform for all kinds of mobile robots, including self-driving cars, delivery trucks, delivery drones and flying cars. Our mission is to empower other companies' mobile robots. As a scientist I work on vehicle-based perception localization systems—a typical robotic system or self-driving car system. The most challenging problem is programming a vehicle to make driving decisions like a human.
“From an engineering perspective, even if the car has perfect perception, perfect localization, it still doesn't know how to drive like a human. What we're trying to do is provide a solution with data generated by autonomous drones. The drones capture high-resolution images and we collect all the data and process it with our algorithm so it can generate beautiful, dense colorized point cloud within one hour. We call it a Dynamic HD Map, which is going to empower multiple components in an autonomous system including perception, localization and the most important part, decision making."
"Decision making is really an umbrella term that includes the driving policy, the motion planning and some part of the control system.”
To access each of these presentations, please click here.