Why automotive software development saves lives?
Over the years, the automotive market has evolved not only towards greater driving pleasure, multimedia comfort features, but also in providing the highest possible level of safety for passengers and driver. Today, all car manufacturing companies speak with one voice: “safety is paramount”. This safety can also be achieved through software that saves lives.
Concise Software has over the years observed the changes that have pushed car manufacturers towards greater levels of safety. Yes, passive systems designed to provide the highest possible level of safety have changed a lot. But as the software development market has evolved, so has the automotive industry. Vehicles are no longer crude, universal machines that operate exactly the same from start to finish. Today a lot can be achieved by updating the software in cars. It is enough that the vehicle has access to the Internet – for example, in the garage. Using the home WiFi network, the vehicle is able to download new portions of software and improve systems already present on board dedicated to passenger safety. Different modules can be updated, from one of the simpler ones at the moment – ABS – to the most complex and sophisticated ones based on sensors and cameras – for example, detecting an oncoming vehicle when driving backwards out of a parking lot. This is one of the more difficult and dangerous maneuvers drivers perform. Software updates are a way not only to improve existing modules, but also to get rid of bugs. As everywhere – in the automotive software development market – there are bugs that, if unnoticed, can cause many dangerous situations.
But software has saved lives before, too. Modules in the car are sets of instructions, now enhanced by developments in the field of artificial intelligence, which operate under specific conditions. We can say that they work today a bit like “fuses” for an imperfect element of the system which is… human. As Tesla’s research shows, most errors are caused by people – not machines.
Reducing the number of road accidents thanks to automotive software development
A 2017 report found that Tesla’s autopilot reduced the number of road accidents for these cars by 40%. It sounds like a number from a fairy tale, but you have to admit that it is very impressive and one would even give credence to the salutary effect of the brand’s cars on the roads. The autonomous car will contribute to reducing the number of accidents, especially fatal ones. Unfortunately, we still have to wait a bit for complex systems, and they must be followed by the entire law and road infrastructure, without which the situation does not necessarily look so colorful. Only then will we be able to talk about fully safe vehicles, whose positive impact can be compared to ABS or airbags. This is one of the most important revolutions to look forward to. However, the emergence of a fully autonomous car requires work in terms of software: creating a perfect system that does not make mistakes, does not hang up and does not cause problems is something to look forward to in the coming years. Such a machine must constantly analyze data from sensors: it must be like a human being who is able to perfectly analyze what his senses are telling him. However, the machine will be so much better because it will not be ruled by bravado, and it will not get tired, it will have better reaction time, and it will definitely outperform humans in terms of how it reacts to changing environments. We mentioned an independent report from 2017 – consider that this was happening five years ago! In that time, Tesla has certainly made huge strides with its autopilot system in terms of vehicles that are capable of self-driving.
The software learns and analyzes the road environment
The technologies that will eventually be used in a self-driving vehicle will be similar to those currently used in cars that are more or less automated. An autonomous car, as the name suggests, performs all driving activities on its own, without requiring driver intervention. Such a vehicle has a number of sensors such as cameras, radars, laser distance meters or accelerometers, imitating the human senses, allowing it to learn and analyze the road environment and what is happening on it. It is no exaggeration to say that the car itself creates a kind of virtual reality. In this way, for example, cameras located in different parts of the vehicle allow processing information about colors and shapes, so they are able to recognize road signs, and radars installed in the car make it possible for the driving vehicle to determine the speed and location of obstacles appearing on its route. It is worth noting that the radars that have been used for years are increasingly being replaced by lidars – devices similar to radars that use laser light instead of radio waves. However, for full autonomy to be possible, the car must not only analyze the information received from the environment, but also process the information received from other vehicles. Communication with cars on the road allows the systems in the vehicles to exchange information about traffic flow, direction of travel or even the road surface, for example. All this information is analyzed by the on-board computer, which not only processes it immediately, but also stores it. In this way it builds up and updates its database, which it can use at any time.
Future-proof software in cars eliminates human mistakes
One of the most important benefits is that with self-driving vehicles, the possibility of human error that could lead to a collision or accident completely disappears – the car will never be tired or distracted by anything. Sensors, cameras and radars placed around the car increase its field of “vision” to 360 °, which eliminates completely the blind spots that drivers have to deal with every day. In addition, the vehicles, well connected to each other and to their surroundings, will allow for optimal route selection adapted to the road conditions, which will have a significant impact on traffic flow, eliminating the problem of traffic jams, among other things, and increasing economy through reduced fuel consumption. Additionally, thanks to self-driving cars everyone will be able to become a “driver” – no one will need a driving license anymore. It all sounds beautiful, but we have to remember that there are no perfect things.
Five degrees of car autonomy
The degree of vehicle autonomy is determined by five levels. Level 0 means no autonomy at all. It includes classic cars in which the driver has full control of the vehicle. Level 1 corresponds to basic driver support in the form of, for example, adaptive cruise control or gentle steering correction when avoiding an obstacle. It is important here to limit the autonomy of the vehicle to only one action for the driver.
In the case of level 2, partial automation is allowed. The car can perform several actions for the driver (for example, in the form of a full parking assistant) with the proviso that the driver must remain alert and ready to take control. Level 3 conditional autonomy allows the driver to place a little more trust in the vehicle. The system can take control of the car when conditions are favorable, such as when driving on a highway. However, if driving in a more complex environment such as a city, the driver is required to take over.
The other two levels are the highest stages of vehicle autonomy. Level 4 assumes minimal driver initiative. The car handles well even in more complex environments. This level allows for no human control of the road, provided that the road conditions are not extremely severe (snowstorm, very heavy rain). The final, fifth level is full vehicle autonomy. Autopilot controls every aspect of the car regardless of weather conditions. In fact, the car can even be without a steering wheel.
A driving school for autonomous cars!
Do you remember your first approaches to the driving test? Huge amounts of knowledge to absorb, details to memorize and learning to react quickly. These same issues are what autonomous vehicles have to face before they are allowed on the road. It may sound funny, but intelligent cars systems actually “learn” road behavior. Currently, this is done through hundreds of thousands of hours spent on the road, during which the cars – under human control – map space and learn to function in an environment of real drivers. The Waabi World project is a road case simulator with enormous capabilities. It aims to enable artificial intelligence to learn the rules of the road and how to react to various incidents that may occur while driving. With this initiative, developers of software for cars will be able to prepare them for potentially dangerous scenarios without having to send the fleet on the road. The company is expected to focus mainly on vans and trucks.
There are already companies on the market working on similar solutions. Software from Waymo or TuSimple is able to simulate even single raindrops and recreate the rays of a setting sun. The head of Waabi, Raquel Urtasun, believes that more important than developing “real world” simulators is teaching artificial intelligence the principles of simple physics. Waabi relies on simulating the response of sensors, cameras and scanners. Waabi engineers want to throw cars under the wheel, literally and figuratively. That’s because the artificial intelligence teaching plan involves testing cases that the algorithm is sure to fail. Urtausn compares the technique to teaching a beginner how to drive, forcing him to follow instructions until he achieves the desired result.
Autonomous cars will, of course, be 100% digital
Every aspect of the roadway, roadside, traffic situation, and road conditions will somehow be recorded, analyzed, and factored into the “decisions” made by the computer controlling the car. That’s millions, maybe even billions of operations every second! People, especially those with more time behind the wheel, don’t realize how many things their brain is analyzing while driving. It is not just watching the lines on the road, or the preceding cars, but also paying attention to the roadside, feeling the whole body behavior of the car, etc. Some of these elements are implemented somewhat subconsciously – like riding a bike, which we don’t really have to think about in order not to fall over. In the case of autonomous cars, all aspects must be taken into account. To make matters worse, road infrastructure has been created by humans for humans – in an analog way – on the principle: “because a human will know that there is a pit to the right of this guardrail”. We know this because we have collected information about it in our brains for a dozen or more years, even decades. The machine has to learn it. That means a gigantic amount of information.
Can software save human lives? Absolutely! Autonomous vehicles promise a much better quality of life for all of us – in an environment where cars drive safely and in an orderly fashion!
You may like to read:
Infotainment system in car – the future of the automotive market
Key features to be considered when developing a car rental app
Different approaches to building car apps
Car safety systems – caring about us every day
Connected cars – What will the car of the future look like?
Car infotainment systems – what’s inside your car’s “brain”?
Why should companies invest in car sharing apps?
Why is software so useful in automotive?