Like a magician setting up a trick, Anuja Sonalker starts by making it clear that there is no hidden driver in her car’s front or back seat. Next, she presses the phone camera up against the side window and waves it around until I reassure her that I’m satisfied. (As with so many tech demos here in 2020, this demonstration is carried out via Zoom, rather than in-person.) Sonalker then turns and strides away from the idling vehicle until she is maybe 10 or 15 feet away. Next, she holds up a smartphone displaying the STEER Tech app and taps it a couple of times.
In the background, the car springs to life. It begins by running a handful of safety checks, making sure that everything from the headlights to the wipers are working properly, and that it has enough fuel to complete its journey. Then it starts to move, rolling slowly into the path of a pair of human-driven cars exiting the parking lot. It stops just in time. When the cars have crossed without incident, Sonalker’s car continues its journey, driving up and down a couple of lanes before it finds a free parking bay, which it finally reverses smoothly into. Mission accomplished.
“Would you like me to summon it for you again?” Sonalker asks, cheerily.
After I answer in the affirmative, Sonalker taps a couple more buttons on the app and the car edges forward, this time pausing to allow a facemask-wearing couple to cross in front of it. Then it repeats the journey in reverse, ultimately ending back in the position it was at the beginning of the demo.
STEER into the future
It’s very impressive. It’s also something that Tesla has been doing for some time with its Smart Summon autonomous parking feature. But there’s a big difference. The technology being developed by Sonalker, founder and CEO of Maryland-based autonomous vehicle company STEER, doesn’t require you to own a top-of-the-range car created by a tech billionaire.
Much like a device such as the plug-and-play Amazon Fire TV sticktransforms your dumb TV into a smart one, this $1,200 box allows customers to retrofit their regular cars with the technology to turn them into self-driving vehicles. Or, at least, self-parking ones.
“What we’re adding is a computational module,” Sonalker told Digital Trends. “It’s a small, square box that houses a mini-computer that runs our algorithm that makes all the decisions. It takes in the [existing sensor inputs of your car], makes decisions, and spends those decisions to the actuators, which then act them out.” (It also uses a stereoscopic vision camera and high precision GPS system for triple-redundancy failure.)
Not every car can be retrofitted with this box, of course. But a surprisingly large number of them can. Basically, if you’ve got a vehicle that was made in the past half-decade or so, there’s a good chance that STEER could help it to, well, steer. The car Sonalker carried out her demo with was a Ford Fusion: A perfectly acceptable runaround, but not necessarily a car that ranks all that highly on the futuristic scale. The tech is also reportedly optimized for the (deep breath) Ford Fusion, Ford Explorer, Ford Edge, Ford Escape, Ford Expedition, Lincoln MKZ, Lincoln MKC, Lexus RX, Chrysler Pacifica, and “in general most vehicles with drive-by-wire capability.” Installing the tech, Sonalker said, takes around half a day.
While STEER’s magic box of tricks isn’t out yet, it is available for pre-order online. Part of that pre-order process involves listing your vehicle so as to confirm that it is a supported model. If it’s not, your car make and model will be added to a spreadsheet so that STEER’s engineers can begin working on the most popular models not yet supported. Brake-by-wire, throttle-by-wire (a.k.a. electronic throttle control), shift-by-wire, steer-by-wire (electronic power-assisted steering) can all be added via additional modules.
Good enough
Software engineers often talk about technology that is “Good Enough.” The principle of Good Enough is that customers will use products that are sufficiently good for their requirements, even if more advanced technology is available. A low-end notebook can’t run a lot of the most compute-intensive software. But it’s cheap and Good Enough for surfing the web or checking your emails. The first Kindles had far worse resolution than a sheet of paper and couldn’t display complicated computer graphics. But they meant that you could take an entire library of books on vacation for approximately the same weight as a supplement-filled glossy magazine. Good Enough goes a long way.
Anuja Sonalker has been working in automotive tech for about a decade. She worked for Battelle, building technology to help protect automotive systems against hackers. Then she worked in a Senior VP role doing the same kind of thing at TowerSec, where her LinkedIn notes that she “focused on cybersecurity for connected and autonomous vehicles, and security challenges of next-generation automotive environments.”
Along the way, Sonalker realized something: that while the dream of fully autonomous Level 5 cars might still be a few years away, existing cars with existing in-built sensor systems are already Good Enough to carry out a surprisingly large number of things. “As I was looking at these systems, I realized that they were already quite powerful,” Sonalker said. “They had a lot of onboard sensors. They had a lot of the electronic actuation that was needed.”
Sonalker thought that there was room in the market for a paradigm that was different from what some other autonomous car companies were doing. “We focused on a very narrow niche of autonomous vehicles, not the whole level five [autonomy idea] which, at the time, the Waymos and Delphis of the world were focused on,” she said. “I felt that you need a very large investment for that, but you also need a lot of time. Developing a fully autonomous vehicle is at least another decade out.”
A different approach to autonomy
STEER’s idea is a lot more humble than that. Its in-car processing unit harnesses the sensors that are already in a car and uses these to carry out an automated valet service. That might be parking in a parking lot or pulling into a parking space on your driveway when you get home after work.
“Very quickly, we were able to build a prototype that demonstrated that our way of doing this, building on top of the existing sensors in a vehicle, was feasible,” she continued. “The cost point was reasonable. The performance was where you needed it to be. And safety-wise it was right for the market.”
On average, upward of 50,000 crashes occur in parking lots and parking garages every year. These result in more than 60,000 injuries and 500 deaths. “They’re very stressful places,” Sonalker said. |People aren’t [always] watching properly. They are looking forward to the end of their journey. I felt that if we could automate the low-speed parking, the endpoint of any journey, it would be a great way to expose consumers to autonomous vehicle technology.”
To be clear, STEER’s technology won’t allow you to be driven autonomously to work any time soon. But once you’ve driven from point A to point B, it will let you hop out of the vehicle, and then find a suitable parking spot, before alerting you via its connected app. At the end of the day, you alert the app once more and your car will start up and drive to a specified location in the lot to pick you up. Sonalker suggested the technology could be useful for seniors worried about parking in busy parking lots, people running multiple errands from home each day, or commuters who, for instance, want to catch a train, who would be able to arrive at the station and then let their car park for them. “Time adds up,” she said. “It’s an asset you can’t get any more of.”
Of course, Good Enough has to be slightly better than cheap e-book readers or low-end laptops when it comes to vehicles. A phone call made over the internet might be Good Enough if it’s free to use, even if it’s occasionally laggy or drops a call in the middle of a conversation.
An autonomous car system, even one that barely exceeds jogging speed, couldn’t make those sorts of mistakes. It only takes a couple of scraped wing mirrors (let alone something far worse) for Good Enough to be Not So Good At All. Sonalker said that STEER’s technology has already proven itself in various safety and efficacy tests. The company has been putting its technology through its paces at BWI-Thurgood Marshall Airport, courtesy of an agreement with the Maryland Department of Transportation.
“You have to work hand in hand with law enforcement, safety officials, fire, all of them, to demo your technology repeatedly,” she said. “You have to show them data for the results… We were the first company in the state of Maryland to actually go through that process and be certified and validated to actually have our cars on public roads in Maryland. So that’s a huge shot in the arm for the technology.”
Can it succeed?
It also remains to be seen whether it will be Good Enough to carve out a “must buy” status for itself in the marketplace. STEER doesn’t have the deep pockets of an automotive giant like Tesla which, as noted, has already developed similar tech at a higher price point.
Will customers be willing to take what they might perceive as a gamble to soup up their cars with this kind of technology? Will STEER manage to broker deals with major car manufacturers to have this technology incorporated into new vehicles as standard, rather than a DIY add-on? All of this will play out in the coming months and years.
One thing’s for sure, though: It would feel pretty great to have an A.I. valet your car to the front of the office so it’s ready and waiting when you set foot outside at 5 p.m. each day. And I wouldn’t miss never having to do another parallel park in another crowded parking lot while other cars line up to wait for me to pull it off. Bring on the robots!
Aftermarket autonomy: This magic box gives your car self-parking abilities [Digitaltrends]