Lessons from Uber: what we learned from the car crash in Arizona

On March 25 in the American city of Tempe, Arizona, there happened a fatal accident, which involved a self-driving car of the company Uber. As a result of the accident 49-year-old Elaine Herzberg died, who was crossing the road in front of the self-driving car.

This accident came as a surprise to many experts. The engineer of automatic control systems, University of California at Berkeley Stephen Shladover, said that automatics should have seen the pedestrian beforehand. The situation, in which the incident with Elaine Herzberg occurred, should not have happened theoretically ― the woman was moving at a right angle to the car, and there were no obstacles on its way hindering the vision.

It is important to note that for the Uber driver, who was in the car at that moment, the appearance of a pedestrian on the road was a surprise ― if the car sensors had to see in the darkness with lidars, the human eye could not see an unexpected obstacle. In addition, the work of a safety driver is different from the work of a usual one ― the attention of a person at the wheel of a self-driving car dissipates, which was shown by the case in Arizona.

The accident of the self-driving car attracted attention of Western mass media and again raised the question of safety and viability of self-driving vehicles. Some experts remarked that this technology is too “raw” at present and requires significant improvements, other experts recalled again the idea that automatics and a living unpredictable person cannot get along on the same road, and therefore separate lanes should be reserved for self-driving cars. Moral questions arose again: how should artificial intelligence behave in difficult, controversial situations? Who should be responsible in case of accidents?

Almost all of the conclusions drawn from the discussion of the incident can be accepted to varying degrees. Indeed, a fully secure transportation system, in which people and self-driving cars operate at the same time, is unlikely at the current level of artificial intelligence. Yes, transport with automatic control should be separated from people whose actions cannot be calculated by artificial intelligence. At the same time, it is necessary to develop intelligent control systems, train neural networks and carry out the necessary research in the field of artificial intelligence to reduce the likelihood of accidents in the future. And SkyWay is already working on it all already now.

The tragic accident has once again confirmed one of the basic ideas of SkyWay transport: to ensure safe movement, a separate way ― a string overpass is required. However even on the overpass, the intelligent system that controls the traffic will have a lot of work ― passengers should be insured both from the collision of modules and from the hit of external objects into the area of running modules. SkyWay developers are creating all these systems and are gradually introducing them into the samples of rolling stock.

The accident in Arizona has caused a lot of negativity in respect of self-driving vehicles, but we must remember that the concept itself, though imperfect, is very promising. A lot of work needs to be done for its improvement, and SkyWay is making its contribution to it right now. We propose to change the approach to infrastructure for the sake of human safety, and time will show that this is the most sound and viable option.

Tags
Trends

Offer publication

Lessons from Uber: what we learned from the car crash in Arizona

Rate and Comment

Recording of the private meeting from November 16 is available

Consent Request Form

Privacy Notice – Newsletter Signup