Problems with autonomous vehicles driving without a driver

Problems with autonomous vehicles driving without a driver

For twenty years now, many large car companies have been working to create a car that will be able to drive without a driver. Even then, they promised that the development of this technology would take ten years, and they managed to develop an application for vehicles that drive in strictly controlled conditions such as warehouses, parking lots and the like. Some vehicles can also drive on the open road, but provided that the driver keeps his hands on the steering wheel so that he can react in case of problems.

The biggest problem with such applications is how to recognize expance and the problems that come from space.

The car app captures space with multiple cameras and motion sensors, and a three-dimensional rotating laser lidar that measures distances in real time very accurately could be used. Ultrasonic sonar could also be used in conditions of poor visibility of swirling dust when even IR cameras do not see anything. All this data in expance needs to be merged into a single space within which the car itself is located on the move.

The space needs to be divided into several levels;

1. soil and objects that are firmly attached to the ground,

2. objects that can move partially in relation to the ground, such as trees, grass, etc.,

3. objects that move freely on the ground such as other cars, people, animals,

4. objects that move freely independently of the ground, such as birds, helicopters, planes, drones,

5. water surfaces.

All such application objects should be recognized and placed at the appropriate level.

At each level, all objects need to be viewed in four dimensions; length, width, height, and time so that a three-dimensional object can be tracked in motion over time. It is necessary to determine the size of the object, the distance of the object, the direction of movement of moving objects, the speed of movement of the object, the possibility of changing the movement of the object in speed and direction.

Today's autonomous vehicle programs pay the most attention to other cars, while assuming that these other vehicles follow traffic rules. Therefore, these programs require a very fast internet connection in order to have constant data on rules such as speed and other restrictions on certain parts of the road.

The problem arises when someone breaks the rules.

And breaking the rules can be for a variety of reasons such as malfunction, carelessness, ignorance, or intent.

Pedestrians and animals can cross the road in many places where this is not allowed, and their movements need to be constantly monitored. The program must recognize such objects that do not follow the rules on the move, and on the basis of all data determine with what maximum speed that a man, child or animal can go out on the road and come in front of the vehicle.

This is a big problem for artificial intelligence because it is based on similar events recorded in large databases, and every time something happens that does not exist in the database the application does not know how to make a decision.

In addition to driving independently, such applications could also be used to protect passengers in the car.

Cameras could recognize the biometric data of all the people who appear near the car. Based on the appearance of the face, pupils, lips, speech or the way of walking, the car could recognize persons who have the right to enter it on their own. He could also compare the people he captures nearby with the dangerous people he has in the database, and if someone like that appears, warn the owner of the vehicle. The application could also spot all people in a sniper position in the vicinity while driving, those who spot a weapon, or those who leave something under or next to the car, and warn passengers about it in real time.

The possibilities of developing and using such cars are great, but it will take a lot of work and time, and it is much more likely that such applications will be used in rail transport rather than on roads, or in off-road vehicles. And when the applications start to be used in railway traffic, then each battery-powered wagon will be able to be independently rearranged in marshalling yards into compositions for a certain direction. Rearrangement will then be much faster, so rail transport will become much faster and cheaper.

And we will have to wait until the car will move on the roads completely independently.

We will have to wait even longer for those who will drive independently and off-road because they will have to assess the terrain itself, ie the hardness of the terrain and the slope, and how flexible the vegetation is on the terrain that the car has to go through.


Other of my technical analyzes and innovations can be found in this book.