It is hard to fathom that any car could drive autonomously without LiDAR and all the necessary sensors. There is one UK startup though, that claims the only components their vehicles need are a GPS and satellite map to operate themselves. The company, Wayve states that not only can they reduce the cost of building autonomous cars, but over time they would actually operate more safely. Wayve cites an approach that utilizes AI to navigate roads with fewer inputs. Wayve employs inexpensive cameras to experience the car’s surroundings that enable it learn how to navigate itself over time. On the surface, this appears to be an excellent potential use of deep learning. The piece that follows provides a lot of interesting detail on the project.
Projects like Google’s Waymo, Uber, Cruise and Aurora are developing autonomous vehicles by throwing engineers at the problem, basing most of their platforms on rule-based systems that try to pre-empt and deal with every edge case, whilst also peppering the cars with more sensors to capture more data. This can work in relatively controlled environments but has the drawback of not being able to flexibly adapt in real-time to fast-changing situations.
Despite all this investment and many years of development, no one has yet been able to launch a commercial autonomous car service. It’s just very hard to hand-engineer. What’s required is not more eyes but better coordination. The simple answer — as it is to almost everything these days — would be to throw AI at the problem, and that’s what many startups, which lack the engineering and hardware muscle of the big players, are trying to do.