• Click to searchClick to close menu
  • Click to view menuClick to close menu

Blog

Main Content Image

Exploring the Speedbumps on the Road to Level-5 Vehicle Autonomy

A stark reality is hitting autonomous vehicle (AV) companies. Whether they’re integrating the most advanced software systems into vehicles, crafting cutting-edge camera technology or retooling the modern vehicle to be the most comfortable haven for commuting, developing safe autonomous driving technology is proving to be harder than previously estimated.

Ford CEO Jim Hackett perhaps said it best in April 2019: "We overestimated the arrival of autonomous vehicles. Its applications will be narrow, what we call geo-fenced, because the problem is so complex.”

Wheels expressed similar sentiments, too.

“We should not expect the technology to be transformative in terms of application or economics immediately,” said Ahsan Rahim, Wheels chief operating officer, in our mobility guide. The comment, which was part of a larger conversation on the outlook of autonomous vehicles, begs the question. While admittedly slow moving, autonomous vehicle technology is assuredly coming.  How should fleet managers stay apprised of its trends while not being led astray by exaggerated predictions?

Protecting the road

What’s the foundational reason manufacturers are in the AV race? Simply put: to save lives. Replacing error-prone human drivers with automated systems that aren’t susceptible to distracted driving will reduce or eliminate the 1.25 million road deaths occurring worldwide each year.

“It's all for the greater good,” said Tim Cengel, manager of manufacturer relations at Wheels. “And it's not even just the deaths. If you think about all the car accidents, the medical costs, the cost of emergency response, insurance—it's a big drain on the modern American.”

If you think about all the car accidents, the medical costs, the cost of emergency response, insurance—it's a big drain on the modern American.
Tim Cengel

To boot, Cengel points to the gains in productivity AVs would allow people to recoup while commuting to and from the workplace.

So, what exactly is the hold up? Let’s dive into the technology that goes into building an autonomous vehicle.

Outfitting the road

Some of the largest players in engineering AV technology are GM and Waymo. GM’s Cruise Automation program is testing in San Francisco, Arizona and a few other areas. They’ve made huge investments and have transformed the Chevy Bolt EV into a fully autonomous vehicle. There’s a rush to get to market, and like many other OEMs, they’ll institute ridesharing to test the technology while recouping some ROI.

Google’s Waymo follows a plug-in AV method. They've purchased 40,000 Chrysler Pacifica hybrids, Jaguars and other vehicles in hopes of retrofitting them into the ideal AV. They too are making large investments with the goal of being first to market.

Ford made news with its lofty deadline of launching a commercial AV by 2021. Now, as mentioned previously, the tone has changed on the short-term feasibility of AVs. And while they haven’t publically adjusted their time table, experts think the roll out may be closer to 2025.

Overall, the challenge remains of how to resolve possibly conflicting information of multiple sensors, as well as the question of how to gracefully handle unexpected situations without needing human input.
Bart Selman

Sensing the road

Lidar sensors are a bit of a gold standard for AV technology. They’re able to scan and detect options near and far, creating a 3D map of sorts. The sensors rotate continuously, generating thousands of laser pulses each second. Through machine learning the data charts the topography, which AVs use to navigate their surroundings.

Yet there’s a contention that believes level-five AV tech is achievable with cameras, radars and machine learning alone. The Lidar-less system would allow vehicles to learn how to navigate complex new environments in a day. This approach, called end-to-end deep learning by Wave.ai, follows a learning model that mimics how humans perceive and adapt to new situations. It follows a reinforcement machine-learning system that allows the vehicle to create rules itself and evolve based on safety driver interventions rather than information manually uploaded by a third-party.

In addition to finding the right sensor, there’s the need for redundancies:

“It is well-known that current computer vision systems can fail in quite unpredictable ways. Having multiple sensors, ideally including Lidar, are therefore critical,” said Bart Selman, a professor of computer science at Cornell’s college of engineering. “Overall, the challenge remains of how to resolve possibly conflicting information of multiple sensors, as well as the question of how to gracefully handle unexpected situations without needing human input.”

Fleet managers—in partnership with their fleet management companies—will need to assess both application and economics to evaluate the strategic opportunities that are unlocked.
Ahsan Rahim

Watching the road

Overall, it’s important to keep a high-level eye on the AV race and consult your fleet management company on the implications for your fleet.

“As the technology develops over the coming years we expect to see the first use of autonomous vehicles in applications like ride-sharing and taxi fleets,” Cengel says. “Fleet managers will have the opportunity to learn from these initial use cases and should begin to think about how their fleet will be changed as the technology becomes available to them.”

Rahim, too, advises as such:  “Fleet managers—in partnership with their fleet management companies—will need to assess both application and economics to evaluate the strategic opportunities that are unlocked.”

In the meantime, we’ll continue to explore the expansive and changing world of AVs. Stay curious by asking questions that will ensure strategic conversations and, ultimately, positive results for your fleet.