Start small

Pete Kelly, Managing Director

12 April 2018

Autonomous Vehicles

Question to Autonomous Vehicle developers: why run tests in big cities first?

The underlying rationale for choosing to do testing and deployment of Autonomous Vehicles (AVs) in large cities seems fairly clear. Such locations offer significant complexity to validate AV technology and high population densities present obvious scale advantages. Other infrastructure – connectivity, charging, presence of mobility service providers, and so on – is also likely to be more advanced than in more rural locations. And, once the technology is proven, it would be a major, and highly visible, milestone along the way to convincing people that AVs can work well.

Developers who have made this choice appear to be deliberately setting the bar high so that, once they have a viable product under those circumstances, the approach can be scaled to many other locations with similar or lower difficulty levels. Unfortunately, it is some of these same apparent development advantages that also appear likely to slow progress in the goal of creating AVs. When the US government committed to putting a man on the moon in the early 1960s, one step was to put somebody into orbit around the Earth – and Russia had already experimented before that by putting a (rather unfortunate) dog into space.

” it would make more sense not to let the best be the enemy of the good”

A more manageable approach would be to develop AVs in an easier setting first. Pre-public testing in campus environments, or in mock-city-neighbourhood locations, has of course already been done. So a small-town, but still real-world, deployment would offer a host of advantages over a large city. Analysis of pre-existing population movements, mapping of environments, nature and norms of traffic flows, and other factors, could be more easily and confidently understood. If extra-vehicular support is needed – such as vehicle-to-vehicle communications, local wireless infrastructure, charging points, live traffic and mapping updates – then creating these systems would be more manageable. Complexity and cost would also be lower as the number of AVs, and other road users, would be smaller. And even though the scale would be vastly reduced, it would still represent a microcosm of the imagined future large-scale city deployments.

If AV developers want to prove that the technology is safe, viable and advantageous, then perhaps it would make more sense not to let the best be the enemy of the good. It is already clear that AV development will take place in step-by-step fashion, with each deployment generating experience and learning for an improved next stage. So why choose difficult and complex locations first and, in doing so, commit to taking some of the biggest steps at the beginning of the journey?

[Full disclosure: the author of this post would quite like to use AVs in the future and lives in a relatively small town.]