The Maturation of the Autonomous Vehicle Space

Reflecting upon the comments and insight from this week’s Automated Vehicle Symposium 2017, the automated vehicle space feels like the Video on Demand space did a couple of decades ago. The technology wizards are making great progress and its obvious the industry is maturing because there are multiple companies addressing both mainstream challenges, as well as corner cases. Still, like video on demand, technology will probably not be the biggest hurdle to autonomous vehicle adoption, but ease-of-use and trust in the technology will probably represent the biggest challenges.

And the importance of trust in the technology by the public is going to be an important factor in how policymakers address automation, according to Dr. Bryan Reimer, Research Scientist, MIT AgeLab and Associate Director of The New England University Transportation Center. He pointed out that the experience people have with Level 2 autonomy will shape their perception about the effectiveness of full autonomy. Reimer’s studies at the MIT AgeLab indicate that vehicle manufacturers still have progress to make.

Reimer brought up a recurring question about the challenges of relying on the driver as a back-up to the computer. To paraphrase what was written on Twitter, how can someone take over the wheel while engrossed in a TV program (or social media, etc.)? His work has led to the idea of a ‘Driver Aware Vehicle’ that senses the driver’s cognitive ability to re-engage, as needed (see this ViodiTV interview for one example of how this could be done).

Regarding the human experience, a presentation by Dr. Edward Steinfeld, Professor at the State University of New York at Buffalo (UB) and Co-Director of RERC on Accessible Public Transportation, recommended that suppliers of autonomous transport “get ahead of regulation by designing for inclusion.” That is, designing mobility solutions that accommodate the entire population, including those that are wheelchair-bound, blind or slowed by age. Design includes all parts of the experience:

• From scheduling to ordering to payment
• Easy ingress and egress from the vehicle
• The ability to easily and safely secure a wheelchair.

The potential of removing the driver and the shift to electric-propulsion with battery packs in a “skateboard”, gives designers freedom not found in traditional ICE (Internal Combustion Engine) vehicles. For instance, the lower platform of an electrified vehicle makes it easier to create ramps with minimal incline for wheelchair access (as opposed to lifts required on traditional paratransit buses).

That one of the vendors in the exhibit hall had a video demonstrating a wheelchair accessible ramp automatically deploying from a vehicle is evidence that the ecosystem is expanding beyond what is essential for autonomous transport and what will be need for commercial viability.

The expansion of the AVS exhibit floor at AVS 2017 was another significant marker in the maturation of the autonomous vehicle space. Where there was once one or two vendors for a particular sensor, at AVS17, there seemed to be multiple vendors; indicative of the makings of a real business.

And the trucking business is a real business that will see efficiency gains and safety improvements from automation. The consensus of a panelists on the topic was that the dire predictions regarding job losses in the trucking industry don’t consider the changing nature of the jobs and the efficiency gains that will help existing truck drivers be more efficient and help counter the current driver shortage, which numbers in the hundreds of thousands (stay tuned for a soon-to-be published interview with industry expert Richard Bishop for more detail on this topic).

As importantly, as mentioned earlier, the push towards electrification and automation will allow for different vehicle form factors (see Einride’s T-Pod, as an example) that allow the right-size vehicle for a given application. For instance, this might mean relatively small, lightweight, electric, driverless pods for intra-city use and larger, driverless tractor/trailers for intercity deliveries.

One of the use-cases that was suggested is that the computer would handle the intercity transit driving on the highway, while a remote driver would take control upon exiting the highway and navigating city streets. An important implication for this type of approach would be the need for a secure, high-speed, low-latency network (e.g. 5G) to allow instant interaction between the truck and driver; a driver who might be positioned a thousand miles away in an air-conditioned office.

The implications of right-sizing vehicles goes beyond the form factor of the vehicle, but impacts the built-environment (e.g. reduced parking demands) and associated infrastructure. One panelist suggested that autonomous vehicles will allow for lighter freight vehicles, which will greatly reduce the structural requirements for bridges and stress on roads (damage increases exponentially with vehicle weight – see this UC Berkeley brief PDF).

Along these lines, it was suggested that transportation planners,

“Follow the model of the Internet and invest in bare pavement. Put the intelligence in the vehicles.”  

The idea is that it will be easier to adapt the end-points (e.g. vehicles) to stay current than to change fixed infrastructure.

Princeton’s Dr. Alain Kornhauser suggested that private entities, such as Lyft and Uber, are effectively public transit solutions (public as available to everyone, but not publicly owned). He asked the important question whether cities and public agencies should continue to be in the transit business, as the cost of mobility drops due to electrification and sharing and private enterprise will increasingly offer compelling mobility services. This is particularly important question if private sharing networks are able to achieve accessibility goals with a much lower need for public monies (here is an example of one such high-level model and how that might work)

One area of consensus was the importance of getting the public dialogue right – back to the trust and acceptance question. Along these lines, a funny anecdote was told about an autonomous vehicle journey where the car adhered to the speed limit laws. Because of this, the car was often perceived as a road boulder and the driver faced occasional angry gestures from frustrated motorists. Understanding this sort of human-machine experience, along with seemingly infinite number of possible human-machine interactions, is sure to have experts returning to AVS next year and for many decades to come.