Level 5 Autonomy: Are We There Yet?

The big question about autonomous cars used to be completely open-ended: What are they?

We answered that question pretty quickly, developing five autonomous driving levels: Driver Assistance, followed by Partial, Conditional, High and ultimately Full Automation — that final futuristic, mind-of-its-own, full-blown robocar you’re thinking of.

Currently, Conditional Automation is on the horizon, with some vehicles leveraging advanced sensor packages, hardware backups and safety software to take control on the road. This massive jump from Level 2 gives drivers freedom to disengage under the correct conditions — think: open freeway. But the next step, the transition from Conditional to High Automation, begins genuine autonomy.

The steering wheel and gas pedal exist for precaution and emergencies — nothing more.

From there, the need for these modules is eliminated altogether, leaving more room within the cockpit for entertainment and productivity. That’s when AVs turn from scary to cool — we’re talking exercise pods, mobile conference rooms and portable theaters.

This concept has expanded the conversation around autonomous cars from “what does it mean?” to “what does it mean for me?” While the timing of fully automated cars isn’t clear, Gartner estimates a quarter of a billion of these vehicles will be on the road by 2020, and analysts expect OEMs to level-up fast or get left behind.

That’s why we’ve started to ask a new question: What are the keys getting this fully automated engine started, so to speak?

Here’s some possibilities:

Over-the-air updates
Being able to update automotive firmware and software is critical for autonomy. Without access to the latest maps and most advanced security patches, a passenger isn’t safe to disengage in the car. As drivers put their trust in manufacturers to develop future-proof vehicles, OEMs need to do more than ensure an outdated map won’t send cars off the road. They need to guarantee a hacker can’t find security vulnerabilities and make the car drive itself off the road.

OEMs need to eliminate cyberattacks as an option altogether. How can they make that happen? By consistently updating software to ensure security against every new hack discovered. Having the capability to do it at the drop of a hat is essential for drivers to embrace driverless cars.

5G connectivity
Bandwidth is the bed-rock of automotive connectivity, and a robust 5G network is the only way to ensure low-to-zero latency data throughput speeds. To have a seamless, and safe, experience within the car, hardware needs to react as fast — or faster — than the human reflex. Vehicle-to-vehicle (V2V) communications will rely on real-time data exchange, ensuring every car receives collision alerts the moment of and react with equal speed.

Real time on 4G network? Not likely.

What else relies on lag-less, expansive 5G? Well, providing those OTA updates that I mentioned. And ensuring cyber-attacks are identified and shut down before completion. Even something as simple as entertainment experiences need to be smooth and seamless no matter where the vehicle is. If we ever want to see the car become a real productivity pod — one with actually instant access to our Hulu account or uninterrupted video conferencing — we need 5G.

Smart city infrastructure
It’s clear supporting a robust, standardized 5G network is the first step a city can take to prepare for autonomous cars. Just as traffic lights and stop signs became standard when vehicles entered the zeitgeist, vehicle-to-infrastructure (V2I) sensors adoption is necessary to guarantee efficient, consistent communications. Exiting a coverage zone could result in omitted map updates or failed V2V communication, putting riders at risk.

But that’s not all cities should think about. 5G-boosted connectivity would allow connected roads to share information from environment to car and track traffic flow to optimize signal cycles and decongest roads via connected stoplights.

A clear monetization strategy
Without clear ROI, encouraging cities to update infrastructure and OEMs to install end-to-end cybersecurity software won’t be easy. Improving the customer experience won’t be reason enough for these parties to invest — especially as ride sharing rises and car ownership declines. What might hit home is that monetizing connected car data could be a $33 billion opportunity by 2025. OEMs and cities alike are seeing it’s time to think outside of the box and inside of the cloud when it comes to making money.

This vehicle-commerce — or v-commerce — model would allow OEMs to leverage in-car data for research studies, helping them see what users like and hate, and make rapid updates. It could even help them deliver relevant, targeted ads via the car’s infotainment system.

The same goes for cities: Understanding traffic flow can lead to optimized display times for advertisements and increased leisure time for citizens, as well as better understanding of where/how people are interacting with their environment.

So when can we expect Level 5 autonomy?

The benefits of these factors are clear, but does that mean we’re close to full autonomy?

Global progress is being made towards driverless cars — an algorithm from MIT allows lane switching with near-human intuition, Stanford developed sensors allowing cars to see hidden objects around corners — something human drivers can’t do — and driverless test modules are on the road in Texas, Belgium, Israel and dozens more regions. The biggest driver, ride-hailing services, might begin piloting driverless services in 2019 and 5G is expected to boom by 2025. But ubiquity is still a decade or two out.

Innovation doesn’t wait for permission, but it might have to wait for infrastructure.

— Stephen L. Surhigh is Vice President and General Manager of Automotive Cloud Services for Harman International. With more than 30 years of operations, sales and management consulting experience, he leads business development for Harman’s Automotive Cloud Services business unit that focuses on data exchange between vehicles and the cloud.