The recent car fatality by self-driving uber has underlined the fact that this technology is not hit up to the mark and feasible for widespread adoption. One of the reasons is there are not many places where self-driving cars can actually drive. Google and other companies have only tested their self-driving cars in major cities where they have spent tedious time ensuring that every lane, off-ramp, curbs, and stop signs are labelled exactly and accurately in 3D. So basically if you live in an area which is unlit, unpaved, or not marked reliably then you can stop dreaming that they will be self-driving cars in your area because such streets are often complicated to plot on the map and have less traffic, there by most companies are unlikely to develop 3D maps for these areas anytime sooner.
One way around this issue is to create systems Advanced enough to make way and navigate without making use of these Maps. Such an effort has been taken by a team of researchers from MIT computer science and artificial intelligence Laboratory where they have developed a framework called MapLite which enables self-driving cars to drive on roads which we have never seen and that too without the use of 3D maps. MapLite makes use of simple GPS data which can be found on Google Maps along with the series of sensors which can observe the road conditions and this allows them to Drive autonomously on unpaved country side roads and also detect roads and it feeds in advance. In the past, such map-less approach has not been used on account of the belief that it is difficult to reach the same level of accuracy and reliability as with detailed and accurate maps.
MapLite is still limited and cannot be relied to drive on mountain roads as it is not yet prepared to understand the dramatic changes in elevation. Therefore, the team is hoping to expand the range of Rhodes on which self driving vehicle can drive.