Cloud over-the-air updates to offer future autonomous vehicles (AVs) refreshed brains
Driverless vehicles will, ultimately, be able to download new ‘brains’ from the cloud and 5G connectivity will hold to key to that capability.
That’s the view of Hemant Sikaria, co-founder and CEO of software specialist in remote vehicle data handling, Sibros. He sees this as being a vital capability if AVs are ever to match the speed of problem solving achieved by the human driver. Sikaria explained: “If you have a truck driver today and the vehicle runs over an object, the truck driver can stop, get out and look under the vehicle. There is no way, even with enough proper sensors, you could figure out exactly what the damage may be and then assess whether you should continue driving or call a repair service.”
The ability to quickly assess a vehicle’s options will require machine learning to reach levels not yet achievable. He said: “So, there are things of that nature that will continue to exist for quite some time. At first you have to assess the information then decide what to do with that. So, as more of these scenarios happen, and the data is recorded, there will be some human intervention to figure out what is the right path. However, that will then get built into the algorithms in the vehicles so the machine learning part of this might improve the data to the cloud, improve the algorithm based on the data you’ve collected from the vehicle. With data from hundreds of thousands of vehicles in the cloud, you can push that back to the car which still functions as it should but sort of gets a new version of its brain every now and then.”
5G connectivity will be crucial in this role, says Sikaria, because of its ability to be predictable in the process of exchanging data with a remote storage facility.
To read the complete article, visit TU-Automotive.