Despite the impression from marketing material, technology challenges remain in making a reality of the fully automated vehicle.
Source: https://datascience-enthusiast.com/DL/Autonomous_driving_Car_detection.html
Whilst the likes of Tesla and Waymo have vehicles which are enabled with automation software, they are not quite ready for large scale open road use. That is not to denigrate the huge progress made – the use of neural networks to perform tasks such as detecting objects and predicting how they will behave would, 10 years ago, have been regarded as science fiction. But even these neural nets need to be trained – through trial and error, in some cases relying on manual labour intensive inputs – whether that is by painstakingly tracing around images on a screen to differentiate a pedestrian from a tree, through to honing in on the most appropriate ‘learning rate’. The technology required to enable higher level of reasoning for complex events in an urban environment remains a challenge yet unresolved.
Source: Self driving car technology: When will the robots hit the road? McKinsey, May 2017 Copyright © 2017 McKinsey & Company. All rights reserved
Of increasing focus are other technology concerns, such as cyber security. The remote 2014 white hat hacking of a Jeep Cherokee led to Chrysler recalling 1.4 million vehicles9. Although this example was part of a research project it highlighted the vulnerabilities in the design of AV’s. The technological loophole is simple: if a hacker can access the internal network of the vehicle, through riding in on the coat tails of a less secure component which is connected to the network, the hacker may be able to gain control of the whole. Given that there are over 100 electronic control units sourced from all over the world, and over 100 million lines of code, it is easy to see why this causes concern. No one has tried a major cyber-attack of this sort, but the risk remains and this unnerves both users and governments10.
Although the technology isn’t fully formed yet, companies and governments are investing large sums in R&D. In 2017 the Brookings Institute estimated that, based on the data collected on 160 deals, investments in the sector approached $80 billion dollars – an estimate which is likely to be an underestimate as it relies on publicly available information. The investment spans the globe and size of businesses, from the large established players such as Apple through to nascent start-ups focused on point to point AV technology solutions. These figures do not include the investment and support being offered by governments – whether that be for testing, R&D projects or related infrastructure investments.
Estimates vary as to how long it will be before the technology challenges are resolved, but it is very plausible that in time these will be overcome. A mid point may be widespread use of less complex AV’s, geo-fenced applications for instance, as described by Jim Hackett, CEO, Ford11.