Page 212 - Are You Future Ready?

To enable this to happen, however the instantaneous
availability of accurate data and its precise interpretation

are crucial.

According to Stanford University professor Sebastian
Thrun, who founded the self-driving car team for

Google, though AV technology is progressing rapidly, a
fully driverless car still faces two major difficulties: the
perception of surroundings and the prediction of upcoming
incidents.

210
A plastic bag or a flying baby?

AVs need to consolidate the information collected from

the surroundings to identify pedestrians, vehicles, road
signs and what is happening, which is challenging for a
computer. Thrun cited one example that in the early days,
it was hard for Google's driverless car to differentiate a

plastic bag from a flying child. Over years, the computer
learnt lessons by combining data from multiple sensors
and the fleet where cars compared sensor readings
with other cars' data gathered on the same road. The