When will autonomous driving finally be possible? Will we be driven by cars in ten years without having to do anything? What else needs to happen?
Ten years is a good time horizon that you can perhaps just about foresee. On the other hand, something can of course happen in ten years that changes everything. But let’s assume a relatively stringent development that is measured against today. If you look back over the period, the race for autonomous driving has only just begun. Ten years ago, Level 3 wasn’t achievable, but it was believed that Level 4 would be available today. Well, Tesla boss Elon Musk believed even then that level 5 would have been reached. He regularly announced that he would soon be ready and today Tesla is still at level 2. Perhaps, as the incident demonstrates promotional video from Tesla, he knew it too and only wanted to act effectively in advertising.
The fact that Tesla lags far behind its own standards is probably also related to the sensors. Tesla does not use lidar, which the vast majority of industry participants consider to be an important basis. The sensors will undoubtedly continue to develop, but distance is what matters most. The further the sensor looks, the faster the car can drive. The various sensors serve as redundancy in case something fails. But they are interesting for different weather conditions, with bad weather still being one of the main problems for sensors. Therefore, most autonomous vehicles are operated in the southern United States. Another sensory problem is the question of cleaning. Because if a sensor is dirty, then perception doesn’t work that well either. That’s why cleaning systems are needed, especially on longer journeys.
However, the interpretation of the sensors also requires a high computing capacity. This has increased significantly in recent years. But it takes a bit more to run the necessary AIs. There is, of course, the option of sending this data to a data center via networking. There, the data set is analyzed more quickly than in the car, but that has data protection problems and lets the car drive into a dead zone. These calculators will do a lot Electricity consume, as a recent MIT study found. That brings us to the next challenge for autonomous driving: the combination of electric cars and autonomous driving – that autonomous e-driving. The batteries not only have to provide energy for the distance, but also for the high computing requirements.
The AIs that perform the navigation and sensor evaluation also need to be improved. This is achieved through training and machine learning, which always generates new ideas. Maybe one day it will be machines that train machines and possibly also build them right away. The question is how far will man trust the machine. That little bit Trustthat exists today has come a long way.
The legal framework was Germany already plugged in, as one never tires of emphasizing. However, the corresponding vehicles do not yet exist and ultimately the question of liability remains in the room. Who is liable if a robot car runs over a human. What would the consequences be? The question of Liability is not specified in the law. There is a massive need for the courts here when the vehicles are on the road. One point reflects the problem: Can one’s own car testify against the drivers? And what if product liability is shifted back and forth between the manufacturer and the software developer?
First and foremost, however, money is needed for development. Lot of money. That was plentiful during the low-interest-rate era. One also speaks of cheap money. But those times are over with inflation. You can also see that in the first wave of start-ups that can no longer finance themselves. Some became bought up, others had to close. The big tech companies, which have had a greater impact on development than the vehicle manufacturers, are currently laying off their workforces.
In ten years, I fear, there will still be no autonomous private vehicles. Perhaps in the extreme luxury segment, but certainly not for the mass market. I would rather guess trains or public transport. Their vehicles wear out more quickly and are more likely to be replaced. The commercial vehicle industry is also being automated in view of the shortage of skilled workers. But probably only on a small scale or maybe only in the US?