At the time the police had explained that it would be difficult to avoid the accident even in the case of a human being driving: Herzberg would suddenly start crossing, suddenly appearing in the middle of the dark road, at a point without pedestrian crossings . After more than a year, however, investigations into the Tempe accident have come to a first set of un-encouraging conclusions for Uber and other companies in the sector: Herzberg was invested due to problems with the software of the Volvo XC90 equipped with systems for driving developed by Uber. According to the US National Transportation Safety Board (NTSB), the US government's investigative agency dealing with transportation-related incidents, the system was unable to identify Herzberg as a pedestrian. However, for the time being the NTSB has not indicated a certain cause for the accident, but will provide further details by the end of November.
However, the experts who conducted the investigation believe that Uber's software had not identified the bicycle as a high-risk collision, until just a few moments before coming into contact with Herzberg, who was carrying it by hand. At that point the braking distance was not enough and the accident occurred. The system was apparently not able to accurately detect an unexpected crossing of the road, in an area without pedestrian crossing and where it was not allowed to pass to pedestrians.
Following a decision taken earlier this year, Uber cannot be indicted for Herzberg's death, but the driver who was on board the SUV and who was due to intervene could still be indicted. From some shots taken by the internal cameras of the SUV it seems that she was distracted, a few moments before the accident, and that she was not observing the road. Other indications indicate that he was probably watching a streaming program on his smartphone.
Uber is just one of the Silicon Valley companies that are developing cars that drive themselves. Waymo, a company controlled by Alphabet (the holding company that controls Google), has been working for years on self-driving cars and has started projects in some cities in the United States to test its systems. Ford and other automotive companies are doing the same, with their own research initiatives or collaborations with Waymo.
Along with the move to electric motors, self-driving cars could be the most important innovation ever in the history of the automotive industry and for this reason billions of dollars of investments have multiplied in recent years. Uber is particularly interested because it considers the transition to autonomous driving essential to make its economic model sustainable: the current one, based on drivers, is not sustainable and is causing huge losses for the company, which has recently been listed on the stock exchange.
Problems and opportunities
Despite the announcements of progress achieved in the last period, cars that drive themselves continue to have some problems in managing all the situations, and the unforeseen events that arise when moving on the street. They are rather reliable on the highway, where there are fewer variables and the most complicated maneuvers are the exits and the passage of lanes, while they struggle in the city traffic where traffic lights, pedestrian crossings, intersections and the presence of pedestrians add different complications for the software that decide how to make move the car.
Waymo, Uber and the others try to emphasize the improvements in autonomous driving systems, while they prefer to leave off doubts and problems that are still unresolved. Before the incident in Arizona and others less serious, these companies tried to promote the safety of their cars as much as possible over the traditional ones.
In 2013, for example, Anthony Levandowski – the engineer behind the project that would become Waymo and later accused by Alphabet of having stolen some documents to take advantage of the creation of his own truck company, he told the New Yorker: "Every year we delay (the transition to autonomous driving) more people die" because of road accidents. Elon Musk, the CEO of Tesla and SpaceX, has often used the same rhetoric, going so far as to claim that anyone who writes articles questions the safety of autonomous driving "kills people".
Net of the positivism that typically pervades the Silicon Valley, it is nonetheless true that under different circumstances the sensors mounted on self-driving cars are more precise than our senses, and above all they allow the vehicle to react to a danger with much lower reaction times of ours. In extreme cases, like the one in Arizona, they fail to remember how a future without investments or deaths on the road is still remote. According to experts, in the case of Tempe probably a human being could not have done better, given the poor visibility and the fact that Herzberg was crossing where it was not allowed.
The security systems that are already there
Developing automatic systems to stop a vehicle from running in an emergency is complicated, and recent tests conducted on the automatic braking systems of some traditional cars indicate how much there is still to work, in view of integrated solutions in the vehicles that drive themselves .
The American Automobile Association (AAA), more or less the US equivalent of our ACI, tested traditional cars, but equipped with emergency braking and pedestrian detection systems, which should offer greater guarantees for drivers and for those walking in the street, reducing investments. For their tests they used mannequins, with a mechanism to make them move and cross a road while a pilot was driving cars to try towards them. The results have been rather disappointing.
In 60 percent of the cases the vehicles hit the mannequins, even in optimal visibility conditions and with a speed of around 30 kilometers per hour. Things got worse in tests with mannequins with dimensions comparable to those of children: investments occurred in 89 percent of cases. Simulating more difficult conditions, including night ones, the researchers achieved even worse results: none of the cars tested were able to detect the presence of a pedestrian on the road at night.
For the tests four cars were used for sale starting this year in the United States and not only: Model 3 (Tesla), Camry (Toyota), Accord (Honda) and Malibu (Chevy). The investment rate reached 100 percent in the scenario where the mannequin crossed the road immediately after a sharp bend, without the car sensors being able to detect its presence.
Some of the automotive companies involved have admitted that their systems do not always work in all situations, due to some technical limitations, but they have nevertheless remembered that in different circumstances they can make the difference, helping to compensate for the slower reaction times of those is driving. Governments and institutions are pushing for incentives for car manufacturers to include automatic emergency braking systems in their vehicles. Last April the European Parliament approved new rules to make them standard by 2022, while in the United States it was the producers themselves who committed themselves to do the same. Meanwhile, different car models are already sold with automatic emergency braking.
Dead on the road
According to the World Health Organization, over 270,000 pedestrians die every year due to road accidents. The deaths of people walking on the streets account for 22 percent of all deaths due to road accidents, with some countries reaching 60 percent. Automatic emergency braking systems can help reduce the problem, but they are not enough if you do not work to improve road safety, ensure safe walkways and areas where cars can only drive at low speeds.
Automatic emergency braking is considered one of the main innovations of recent years with regard to active safety systems, although as we have seen the specific ones to recognize and avoid pedestrians are not always effective. Some manufacturers have started to provide these solutions as standard, while in the meantime they have dedicated themselves to developing other safety systems such as: sensors to automatically keep driving in a lane, cruise control that automatically adapts to road limits and traffic, self-parking.
Autonomous driving levels and safety
The prototypes of self-driving cars already use all these solutions and integrate others, even more sophisticated, to detect what they have around, to distinguish fixed objects from moving objects, to recognize horizontal and vertical signs, to automatically follow paths without the human intervention. The combination of all these systems offers greater guarantees with regard to safety, but the level of reliability varies greatly and only in the last few years with the improvement of algorithms and artificial intelligence has there been significant progress.
To bring order to the various classifications of autonomous vehicles, in 2014 SAE International, an institution that coordinates the rules in the automotive sector, identified 6 different levels for automatic driving, based on how much a human being must intervene to compensate for the choices of the machine: level 0 is that of most cars in circulation, while 5 is the one in which the vehicles will have reached complete automation.
Although there are exceptions and the classification is often debated by individual producers, we can say that today most self-driving cars are between levels 2 and 3.
A full level 3 requires that, under normal conditions, the car manages all the main aspects of driving, choosing direction and accelerating and braking. The presence of the driver is however necessary to resolve situations in which the car's software shows uncertainties or if conditions occur that make driving safely safe. With the option "Autopilot" active, the Tesla currently on sale meet part of the requirements identified in level 3 and, on several occasions, have been shown to be able to avoid accidents with response times much faster than those of humans.
Overall, the risk of accidents and investments will tend to be reduced as the last two levels of classification are reached, but it will not be easy to get there. The 4 in fact provides that the car is able to handle any eventuality and unforeseen, except for extreme conditions determined for example by bad weather. Level 5 will be reached when the cars can run completely on their own, thus making it possible for them to move without anyone on board to monitor their driving.
The current safety assessment of autonomous systems compared to traditional ones is carried out on a statistical basis, calculating the number of fatal accidents per kilometers traveled. Traditional cars are involved in about 1.2 deaths every 160 million kilometers; Waymo's cars had traveled 8 million kilometers in February 2018.
The only fatal accident involving a known level 3 car has so far been the one in Arizona last year. Five other accidents involved as many Tesla with active autopilot: in all accidents the driver died due to his injuries. According to Tesla the accidents would have occurred even in the absence of the autopilot, due to the external causes that caused them. Tesla is the only automotive company to already sell an advanced automatic system, and consequently it is normal for it to register more accidents, covering a much greater number of kilometers of companies like Waymo still with vehicles being tested.
To date, level 5 seems to be still a long way off, but according to the most optimistic the target could be reached relatively quickly. A push forward has been provided by the evolution of artificial intelligence systems, which have become more accessible thanks to the greater computing power of the processors and to models optimized to manage them, even with limited resources. The large amount of data collected from prototypes already on the road for years, or from models already on sale like Tesla, offers further opportunities to integrate the new systems and make them "smarter" and safer.