Ico Maker - stock.adobe.com

Autonomous vehicle tech is too costly and unreliable for production

Level 5, fully autonomous vehicles are a decade away. By then, the tech will be so expensive, that it will only be available on high-end vehicles

A study by analyst Forrester for chipmaker Arm, looking at the development of autonomous vehicles (AV), has reported that engineers believe the cost of the technology needed to support autonomous vehicles is too high.

Forrester’s study found most of the organisations that have developed AV prototypes are struggling to translate these into safe, secure and affordable production-level designs.

When asked about when fully autonomous vehicles would be available, one machine learning and perception engineer interviewed by Forrester said: “Production-ready vehicles will be more than 10 years [out]. And they will definitely only be in premium cars for a long time after that.”

Forrester’s online surveys of 54 global AV practitioners found that 22% believe component costs are too high.

Reliability of the systems is also a big concern. Over a third of the engineers who took part in the study said they are concerned with software that is not behaving acceptably in universal situations, the high cost of components and securing the vehicle systems from cyber attacks.

The Society for Automotive Engineers (SAE) puts the evolution of self-driving vehicles into five categories: from Level 1, which is totally manual, to Level 5, where the vehicle is fully capable of driving itself without human input.

Level 2 provides some degree of driver assistance, while at Level 3, vehicles can steer, accelerate or decelerate, and pass other cars without human input. Level 4 is an advanced stage of augmented artificial intelligence (AI), where the vehicle will slow down, pull over or park itself at a safe spot if the driver fails to take control when requested.

Read more about autonomous vehicle technology

  • For autonomous vehicles to drive themselves in the future, it is critical to collected connected car data. Esurance's Arun Ganesan outlines what data is collected and why, and how it will help.
  • We talk to AI experts about the challenges of teaching autonomous vehicles how to drive safely in often extreme environments, and how deep learning simulations can help. 

While their goal may be to reach Level 5, many of the engineers Forrester interviewed said the sensor technology needed for production-ready AV systems is still nascent.

One director of automated driving programs told Forrester the challenge in going from Level 4 to Level 5 automation is that the sensor technology does not exist. “Computing horsepower and algorithms don’t exist yet.

The amount of human judgement in driving in boundless conditions is such that it will take over 10 years before a computer can make all of the judgments in a guaranteed safer way than a human in all situations. The computer is going to have to be at least a factor of 10 better than what a human can do before the public will accept it.”

In a blog post about the study, Robert Day, director of automotive solutions and platforms at Arm, said: “All of my autonomous vehicle maker contacts tell me the current banks of power-hungry computers fed by arrays of expensive sensors won’t scale. And despite the huge processing power deployed, these systems still require backup human drivers.”

He said Arm is positioning itself to support AV in production, by focusing on efficient AI that is both secure and low cost.

Next Steps

Find out how DXC Technology supports BMW's autonomous vehicle development

Autonomous vehicle technology advancing amid big challenges

Read more on Software development tools

CIO
Security
Networking
Data Center
Data Management
Close