CMU Red Team Vehicle for the 2004 DARPA Grand Challenge

The Week in Mobility — 25 June 2021

Self-driving Long Game

In 2006, UK-based game company Mind Candy tested the theory of six degrees of separation as part of an Alternate Reality Game (ARG). They provided a photograph of a man, a name, and the Japanese characters that translate to “Find me”. The identity of Satoshi, the man in question, was revealed in 2020 as advancements in facial recognition coupled with a life migrated online meant the 14 year search could finally be concluded in a matter of seconds.

In 2006 this would have been inconceivable. At that time the term “facebook” would have been interpreted by most as an odd description of a photo album. Despite that, the internet had existed since 1990 and facial recognition was first pioneered in 1964.

In short it took 56 years for the right concoction of coalescing technologies and behavioural shifts for the search for Satoshi to be both easy and importantly — not novel. For autonomous vehicles we are experiencing something a little similar, patiently waiting for the moment when self-driving is a ubiquitous phenomenon.

2004 marked the first DARPA Grand Challenge, which was created to spur the development of fully autonomous ground vehicles, held its first competition on the Mojave Desert along a 150 mile route. None of the vehicles completed the challenge. Fast forward to 2014 and ACES (autonomous, connected, electric and shared) made its way into mainstream automotive lexicon for the first time. One year later Daimler announced its self-driving truck which was cleared to test on US highways and since then we’ve been touting the same safety and human error improvements that autonomy promises.

6 years later and we are somewhere between some very substantial ADAS features (Advanced Driver Assistance Systems) and deep disappointment in the lack of robo-taxi that, knowing precisely the time between finishing brushing our teeth and leaving the house, automatically arrives outside the front door.

Waymo, Cruise and Zoox have led the charge with many others coming and going — including Uber and Lyft’s self-driving subsidiaries. But even the self-driving superstars have not been without criticism of their progress or scepticism over the staggering amounts of cash they’re consuming. Waymo recently raised $2.5 billion in its second external investment round and GM has just given a $5 billion credit line to Cruise, forcing some to ask if self-driving is even a justifiable expense.

Meanwhile Tesla is aiming for autonomy that is not 3D vision based (ie no Lidars). Instead gathering data from the vehicle cameras (2D data) from its vehicles on the road to teach its self-driving systems. Processing this data requires the 5th largest supercomputer in the world despite Tesla vehicles representing just 0.07% of the world’s driving.

17 years on from the first DARPA Grand Challenge and there is still a lot of work to be done. The advancements required in neural networks, regulatory environment and public acceptance mean we won’t get a sleeping pod to work anytime soon. However, reducing accidents is imperative and with human error responsible for roughly 90% of traffic accidents — driverless vehicles have been heralded as the shining beacon to save us from our dangerous, distracted driving ways.

But removing the human need not be the only solution to eliminate the 1.35 million annual fatalities on the road. Reducing VMT (vehicle miles travelled), greater bus ridership, cycle lane and sidewalk provision, data gathering and analysis at intersections and spreading commutes throughout the day are examples that would all bring substantial improvements to safety. The trouble is, none of these sound very exciting. We are more fascinated by things that surpass human ability than changing our behaviours. Is it laziness or a burgeoning expectation that technology will solve all problems? I’m sure there are many people who would rather wait for carbon capture tech than give up single use plastic.

I think we need to lay off the “Self-driving cars won’t be here any time soon” nay-sayer narrative and instead remember that their motivation is not novelty or a quick win, but safety. In the interim, we must shift our focus to the technologically limited, freely available human choices we can make today to reduce accidents; ensuring that when self-driving vehicles do arrive they will make an incremental improvement to safety rather than the step change we need them to do today.

Elsewhere in the industry

  • 7 Eleven is adding DC chargers at its stores in North America —
  • The City of Manchester are building a multi-functional mobility hub — Cities Today
  • Jaguar Land Rover will trial an hydrogen Defender later this year — AutoCar
  • Aurrigo’s ugly autonomous shuttle scares passengers away — NextWeb
  • Multi Brand car workshop platform, GoMechanic raises $35 Million — EnTrackr
  • Indian logistics giant, Delhivery raises $277 million ahead of IPO — TechCrunch
  • Audi will be EV only by 2033 — inputmag
  • Medical transportation startup Ride Health receives $10M in funding -mobihealthnews
  • First Bus are transforming Caledonia Bus Depot into an EV charging hub — IntelligentTransport
  • BYD reveals its electric D-type school bus — TechTimes
  • Circulor, a supply chain tracing company raised $14 in funding — FreightWaves
  • Worthy of your time: The China Association of Motor Manufacturers has made a repository for manufacturers to purchase self-driving data — Bloomberg

Fully Charged, ex-Arrival Ltd —Sustainability, Mobility, Tech, Books and anything in between