Insights on Transportation: A Shift to Autonomous Vehicles

By Andrew Der

According to Quartz, in California, Texas, Arizona, Washington, Pennsylvania and Michigan cars are already driving themselves. However, there are numerous challenges up ahead before autonomous vehicles become mainstream. From government regulations to additional data, it could be a while before we see this technology hit the road. On the other hand, it could be sooner than we realize. Wondering what opinions are out there? Four portfolio companies share their thoughts on when we’ll see autonomous cars hit the freeway. In this month’s “Insights On” blog post, I asked four of our portfolio companies, Urgent.ly, Veniam, Renovo and Swiftmile to share their thoughts on when we’ll see autonomous cars hit the freeway.

What do you think the biggest factors are preventing autonomous driving?

Lokesh Kumar, Co-founder and VP of Tech, Urgent.ly: In the short — term, the biggest factors preventing the mass adoption of autonomous vehicles are the lack of Federal Regulations around liability coverage, software and mechanical hurdles; resistance to change; and probably most important, human error.

Humans can make mistakes while driving, including abrupt braking, lane veering and changing, presenting a difficult task for autonomous cars, which are hard-wired to perform a certain way under given circumstances.

João Barros, Co-founder and CEO, Veniam: There is still quite a bit of work in getting vehicles to drive autonomously, not just in highways and flat suburbs, but also in high-density urban areas that are more dynamic and less predictable. A major factor that is largely overlooked is the need for vehicles to share massive amounts of data between each other and the cloud, for example large high-definition maps and software updates.The vehicles will need to learn how to talk with each other with very low latency, not only to avoid accidents, but also for platooning and collaborative navigation. Current networks are not ready to support this traffic.

In addition, new policies are needed to accelerate adoption of mobility services delivered by fleets of autonomous vehicles. Rather than forcing autonomous cars to share space with human driven cars, I believe the easiest path is to define districts that are reserved for autonomous vehicles and grow their city footprint from there. It is likely however that there will be an extended period of mixed-use (driver and driverless) vehicles. This will necessitate more data offload as cameras and other sensors will become the eyewitnesses to events.

Chris Heiser, Co-founder and CEO, Renovo: Highly automated vehicles (HAVs) need data to train their algorithms, and this is a significant challenge for the industry. These “training sets” of data contain the large number of rare edge cases that HAVs need to learn to ensure they can operate safely under all conditions. However, building these sets of data is cumbersome, expensive, and takes a lot of time.

Simulation tools are helping to close the gap, but data sharing can accelerate the market in a major way.

Leo Vera, COO-CFO, Swiftmile: Infrastructure, city design, safety, and scale will continue to be the biggest challenges facing the autonomous vehicle revolution.

When will autonomous cars be on the road and why?

LK: Vehicle-to-vehicle as well as reliable vehicle-to-infrastructure communication will become imperative as intelligent transportation permeates the market, allowing autonomous vehicles to learn from other cars and their environment in real-time.

JB: We are witnessing the start of a revolution that touches all players of the transportation ecosystem, including auto OEMs, insurers, municipalities, telecom operators, and mobility service providers, to name just a few. Given the economic interests and the billions of dollars of investment flowing into this space, I believe the transition may come sooner than many think. The predictions of incumbents have often been off by many years, case in point telco OEMs with respect to packet switching or headset manufacturers with respect to the smartphone penetration.

My estimate is that we will see autonomous vehicles appear widely in campuses and industrial environments in as early as 2020 and grow significantly beyond those confined spaces by 2030. The stars are aligning for a major technological transition.

CH: Simulation tools and data sharing will lead to HAVs on the road — without safety drivers — in trials by next year. A larger challenge is deploying a scalable business model that is safe and cost-efficient.

LV: We believe eBikes will be the proving grounds for critical V-to-V and V-to-I communications in the next three to five years, thus laying the foundation for autonomous cars.

What needs to be solved before autonomous cars are on the road?

LK: The AI implications here are reflective of the critical need for autonomous cars to adapt to changes in transportation norms over time, as more autonomous and connected solutions are integrated into the traffic grid.

JB: We need strong AI for unpredictable environments, vehicle mesh networks that can support massive data flows and a legal framework that fosters the transition from early pilots by early adopters to full scale commercial deployments, initially in confined districts that can be expanded to entire cities.

CH: We believe the right model for HAVs is Automated Mobility on Demand (AMoD) in our urban centers. AMoD requires car companies, fleets, suppliers, networks, and AI providers to collaborate. The AMoD approach needs open, interoperable software ecosystems that are safe and secure — this is what we focus on at Renovo.

LV: Before autonomous cars take the road, light electric vehicles such as electric bikes will need to pave the way. Through light electric vehicles, smart city grids and communications infrastructure we will develop on a low-risk, safe, and scalable way in order to prove out the necessary systems for autonomous cars to truly work.

There are still a few challenges remaining before autonomous vehicles are massively adopted. While we are currently witnessing the start of the revolution, training vehicles with data to operate in the same way humans drive is still a big hurdle to overcome. Additionally, federal regulations stand in the way of autonomous cars being on the road. Time and advancements in technology will tell when autonomous cars are ready for the market!

Read more from Verizon Ventures: