09 Oct Driverless Car Accident Reports Make Unhappy Reading For Humans
As technology giants accelerate humanity towards a driverless car future, where we are conditioned to keep our eyeballs on our devices while algorithms take the wheel and navigate the vagaries of the open road, safety questions crash headlong into ethical and philosophical considerations.
Earlier this year Google blogged about the eleven “minor accidents” its driverless cars had been involved in over six years of testing — laying the blame for all 11 incidents at the hands of the other human drivers. Which sounds great for the technology on the surface. But in reality it underlines the inherent complexities of blending two very different styles of driving — and suggests that robot cars might actually be too cautious and careful.
Combine that cautious, by-the-book approach with human drivers’ tendency to take risks and cut corners, and well, that, in itself, might indicate driverless cars’ risk aversion is an accident waiting to happen (at least when human drivers are also in the mix).
Google is now trying to train its cars to drive “a bit more humanistically”, as a Google driverless car bod put it this summer, using a word that seems better suited to the lexicon of a robot. Which boils down to getting robots to act a bit more aggressively at the wheel. Truly these are strange days.
Autonomous vehicles navigating open roads guided only by algorithmic smarts is certainly an impressive technical achievement. But successfully integrating such driverless vehicles into the organic, reactive chaos of (for now) human-motorist dominated roads will be an even more impressive achievement — and we’re not there yet. Frankly the technical progress achieved thus far, by Google and others in this field, may prove the far easier portion of what remains a very complex problem.
The last mile of driverless cars is going to require an awful lot of engineering sweat, and regulatory and society accord about acceptable levels of risk (including very sizable risks to a whole swathe of human employment). Self-driving car-makers accepting blanket liability for accidents is one way the companies involved are trying to accelerate the market.
As you’d expect, California has been at the forefront of fueling tech developments here. Its DMV is currently developing regulations for what it dryly dubs the “post-testing deployment of autonomous vehicles” — a process that’s, unsurprisingly given the aforementioned complexities, lagging far behind schedule, with no draft rules published yet, despite them being slated to arrive at the start of this year.