Self-driving cars’ s are largely tested in cities and that is no coincidence. You might think that there are more interesting traffic situations and therefore it is useful, but the predetermined roads, sidewalks and other obstacles of a city are much needed for most self-driving systems in the form of a 3D map.
Finding the way
MapLite is basically not very different from other autonomous systems: the car gets an idea of where it is via GPS and from there an end goal is generated and a ‘visible’ for the car. local purpose. After that it will be different: LiDAR will look at where the edges of the road are by looking where the road ends with flatness. In this way a route (or in any case a road) can be followed without a map.
In addition, general parameters are used to make the car make decisions at intersections, oncoming traffic and other separate situations. In this way the system can handle a lot of country roads, but there are still some gaps in the knowledge. MapLite, for example, still has a lot of trouble with mountain roads and other routes with significant differences in height, but the researchers want to eliminate these imperfections and then there should be a system that should actually be able to run on every deserted road in the world.
That is quite a breakthrough, because although it is nice that there is so much data available for self-driving cars, the possibility of improvising is even more important. If only to make sure that if a joker moves a post, autonomous cars do not massively make the wrong guess. Not only that: the less a car depends on a pre-programmed system, the less vulnerable it is to any hackers. In this child phase of technology that is not such a problem, but if it is really going to be put into use, every bit less dependency is more than welcome.