Autonomous Cars – End of the Year or Not Yet Near?

Mark Patrick
Autonomous Cars – End of the Year or Not Yet Near?
Tesla CEO Elon Musk has predicted that his company’s cars might be able to drive across the US autonomously as early as the beginning of 2018. Is this ambitious goal possible?

Before we examine that question, though, what do we mean by “autonomous”? The most advanced self-driving vehicles on the roads today can drive themselves for extended periods, but still need human oversight, so it’s debatable whether they can simply be described as autonomous. We need a more fine-grained description.

The US Department of Transport has adopted a standard created by the Society of Automotive Engineers (SAE). The standard, J3016, defines six levels of autonomy, from level 0, which means there’s no autonomy and the human driver is in full control, up to level 5, in which the vehicle can drive itself anywhere, and a human is optional.

Under the J3016 standard, level 1 means the vehicle uses a driver assistance feature such as cruise control or lane keeping. Level 2 is similar, but the vehicle may use multiple driver assistance features simultaneously. Level 3 is the first level at which the driver can safely take his or her hands off the wheel, though only in certain driving environments – the driver must still monitor the road and be ready to retake control at any time. The most advanced commercially available cars, such as Tesla models, are generally seen as being at level 3.

At level 4, the vehicle remains safe even without human monitoring or intervention, but only in certain clearly bounded scenarios – for example, freeway driving in good weather. Finally, at level 5 the system performs at least as well as a competent human driver, in all road, traffic and weather situations.

Fully Autonomous Driving by 2018?

The fully autonomous cross-country driving predicted by Musk would require at least level 4, and probably level 5, autonomy by early 2018. What do other developers think?

The autonomous vehicle industry is “not even close to level 5,” Toyota Research Institute CEO Gill Pratt says. “It’ll take many years and many more miles, in simulated and real-world testing, to achieve the perfection required for level 5 autonomy.” Even with huge advances in technology, “people have zero tolerance for deaths caused by a machine,” Pratt added. Toyota’s research is currently focused on level 2 autonomy.

BMW researchers believe level 4 autonomy is about 10 years away, according to Dr. Dirk Wisselmann, Senior Engineer of Automated Driving at BMW. BMW’s autonomous vehicles are currently at level 2 and the company is working on level 3.

Google, one of the most prominent and technologically advanced self-driving vehicle developers, is more optimistic. Waymo, an independent subsidiary of Alphabet, Google’s parent corporation, is developing level 4 or level 5 autonomy, according to chief executive John Krafcik. Prior to spinning off Waymo, Google had aimed to begin commercializing the technology by 2020. The company has apparently abandoned plans to sell cars itself, but is cooperating closely with Fiat Chrysler and Honda, and is also prepared to license technology to others.

Autonomous Driving’s Uncanny Valley

What’s behind the big difference between the most optimistic and most pessimistic predictions? Some researchers believe there’s a deep pothole on the road to full autonomy, and it lies at level 3. In fact, some want to go directly from level 2 to level 4. Representatives of Google’s Waymo, Ford, and Toyota agree level 3 may even increase risk. “There’s evidence to suggest that level 3 may show an increase in traffic crashes,” Nidhi Kalra, of the Rand Center for Decision Making Under Uncertainty, told a US congressional hearing.

The paradox of level 3 autonomy, according to skeptics, is that it demands drivers be ready to take over swiftly in emergencies, but undermines that readiness by creating an environment so undemanding that drivers are lulled into a false sense of security. Ford denies reports of its test drivers falling asleep behind the wheel, but “it’s human nature that you start trusting the vehicle more and more and feel you don’t need to be paying attention,” Ford CTO Raj Nair admitted.

Indeed, Tesla has seen fatal accidents in which its cars ran into an obstruction while the driver was apparently not paying attention.

Despite this, Musk still believes that level 3 autonomy is not a hurdle, but a critical stepping-stone on the way to full autonomy. He appears to argue that it may eventually provide an overall reduction in accidents: “As soon as you have data that says that autonomy improves safety... it would be morally wrong to withhold functionalities that improve safety simply in order to avoid criticisms or for fear of being involved in lawsuits.”

However, even the ebullient Tesla CEO admits that his prediction of level 4 or 5 autonomy by 2018 “might be slightly optimistic,” and, as we’ve seen, other industry figures seem to agree, so a date well into the 2020s is seen as most probable.

Share this page


Want more like this? Register for our newsletter
developing bluetooth BLE beacons Joe Tillison | Silicon Labs
Determining BLE Beacon Proximity is a Challenge. Here's How It's Done Today
Bluetooth is an ideal wireless technology. It has developed over the years and as a result it is being used in many applications, although in some cases it presents some interesting challenges.
Training
Online - RF and Wireless Essentials for Engineers
This on-line course enables you to quickly get up-to-speed & understand key concepts of modern radio frequency, RF & wireless communications systems

More training courses

Whitepapers
Using Digital Control Designs for Stable Power Supplies
Find out how to achieve stable power supply designs with fast transient response by using digital control techniques in this whitepaper from Intersil.

More whitepapers