The race for fully autonomous vehicles is, as we know it, fiercer than ever. From tech giants such as Baidu, Delphi and Google to otherwise conventional automakers such as KIA and Ford. There are actually too many companies to name who have in recent years joined the cause for self-driving cars. We have seen quite a lot of footage of Google’s driverless pod-like vehicle roaming around Silicon Valley and other parts of the state including a hilarious encounter with a police officer who pulled the car over for driving too slow.
A familiar trend that very few people noticed was all of the data that these autonomous vehicles had collected was in broad daylight and in ideal sunny conditions. This is because the self-driving cars use a light and range detection system called Lidar. The system works by firing laser light away from the vehicle and measures how much light reflects back much like radar uses radio waves.
However, the system’s functionality is fatally hindered by harsher weather conditions, when the road markings aren’t clear enough for the car’s on-board camera to see. Ford Motor Company (NYSE:F) recognizes this issue and has announced that it will test its concept autonomous vehicles in snowy conditions at the Mcity test facility in Ann Arbor, Michigan.
The main idea is that just as human drivers subconsciously use various landmarks to determine their location on the road in uncertain situations, computers would have to learn how to do the same thing if they are to remain functional in harsh weather conditions such as rain, snow or even just road salt on the sensors.
As GPS is too inaccurate to tell the car’s precise position at a certain time, Ford’s solution instead involves programming the Lidar sensors to pick up on landmarks above the ground, such as road signs; to then compare with an existing high resolution roadmap of the area that would in turn compare the data with information gathered by another autonomous vehicle that has taken the same route in more favorable conditions.
Entities such as TomTom and Nokia’s HERE maps are excessively working on not only collecting this data but also how the data will be processed so that the autonomous vehicle can determine its exact location on the road so that storage and data processing capabilities are optimized. TomTom are reported to have already started to work on a similar process called Road DNA.
All of this makes sense because if autonomous vehicles are to come into our lives, they should come as an aid rather than an unwanted piece of technology; it should learn to take over control in critical situations where the driver is unable to prevent a collision rather than the other way around. Autonomous vehicles in an ideal world would allow the human as much control as is safe, but for that to happen, it is essential that first we clearly define what the boundaries of safety and risk can be.
Without such tests, autonomous technology will just become an intriguing little thing you let your children use on a Sunday to let them pretend that they’re the ones in control of the car. Various sensor suppliers such as Quanergy systems and Continental are making some serious headway in this very matter and very soon we could see the developers of autonomous technology move on to the next hurdle in the path to fully autonomous driving vehicles.