This weekend was a big reveal for Ford, which showed off its new GT super-car. The quarter of a million pound vehicle has already begun shipping in China and has been well received, but isn’t expected to appear on Western shores until sometime next year. To celebrate the continued roll out of the car, Ford performed a light show this weekend, showing the car’s development from concept right through to its final version. However, the big news for us wasn’t the new GT, but that Moray Callum, Ford’s Vice President of design, also stated that the company’s driverless vehicles were around 90 per cent complete.
This is a technology that Ford has been working on for almost two decades he said, suggesting that internal developments had gone through many stages over the years. However, it had finally reached a stage of near completion, with Callum suggesting that Ford’s vehicles can drive in almost any scenario on any road in the world, it was just a case now of making them capable of doing so when certain situations arose; like accidents, road works and other random obstacles that are difficult to predict.
This is the same sort of stage of development that we’ve heard other manufacturers are reaching. It’s something that’s hard to program for, since there are many different factors to take into consideration and while humans can react instinctively to it and continue to evolve their actions as the situation changes, that’s much harder for an AI to do.
Some companies have developed systems whereby an autonomous car will sit in traffic, or wait for a roadblock to clear for a set period of time and if nothing changes by the end of that countdown, it will either turn around and attempt to find an alternative route, or will request input and direction from an on board ‘passenger’ or failing that, get in touch with cloud servers that can give it more information to work with.
Those sorts of outside influences may be what certain automated vehicles need while the technology is perfected, as they will surely come into play often in the first generation of driverless cars. However, it’s unlikely to be too long before they’re figured out, as once we have real world scenarios to use in the creation of future algorithms, there will be precedent that can be set for the AI, something that makes it much easier to program for.