An Introduction to Autonomous Vehicles

Have you ever thought about being in traffic back home and rest in your car by taking your hands off the steering wheel? Or read a book, watch videos while the car is in the chaos of traffic jams?

Well, know that this possibility is already being developed and may soon be available.

I have many tech passions, but among the latest transportation industry trends, autonomous vehicles' development is undoubtedly one of the most intriguing for me.

While this innovation seems to usher in a new era in the transportation market, I’ve been dedicating a lot of time studying and developing a series of “home-made” solutions for autonomous driving that better understand this fantastic technology.

Similarly, it is interesting to know in-depth about how the current self-driving technology works.

Since electronic injection, many automobiles have changed, and other systems have been incorporated into the ECU (central module).

Nowadays, there are airbag control modules, ABS brakes, stability control, autopilot, and immobilizers.

These electronic controllers will need to make a giant leap in quality, efficiency, and safety in autonomous cars.

We will be driving in a better world.

I firmly believe that the impact on traffic safety will be the great legacy of autonomous cars.

Accidents involving cars are the 8th leading cause of death globally, and 95% of accidents are caused by human error.

Bearing in mind that car accidents are the 8th leading cause of death globally, and 95% of the accidents are caused by human error, the expectation is that transport automation will represent a significant reduction in the number of occurrences and, mainly, of victims.

Learning by doing: Home-made autonomous driving systems

The higher the level of automation of a vehicle, the greater the influence of AI and computing on its components. Therefore, I am fascinated to understand the logic that drives these machines attractive, both concerning hardware and software.

Understanding the concepts of electronics and Artificial Intelligence applied to autonomous vehicles has become one of my goals during this period.

I’ve started with the classic “smart RC cars,” coding in python, and adding some necessary sensors to understand how to control small electric motors and servos. Now I’m doing something a little bigger.

I’m working on building an autonomous car for my daughter… well… she is three years old..so it will not be a Tesla Model Y with autopilot. It will still be a “power wheels” ride-on car with level 2 of automation — (see below for Automation Levels). It will use AI and a series of sensors that currently I’m applying to it…

I will write a dedicated article explaining this new project with more details very soon, but before that, let’s review together with the basics of this unique technology:

What are autonomous vehicles?

Autonomous vehicles are cars (but also trucks, buses, and others…) where human drivers are not required to take control to safely operate the vehicle, combining sensors and software to control, navigate, and drive the vehicle.

To allow self-driving vehicles to take over the streets, there is a trend by automakers to gradually add technologies that collaborate with drivers both in driving and maintaining the car itself, such as:

• cruise speed control;
• driving and parking assistance;
• braking management;
• obstacle detection system and road users;
• proximity alerts with other vehicles and driving adaptations;
• monitoring of operating conditions;
• speed adjustment according to path conditions.

These are some of the systems already present in specific models. To better understand how integration occurs, let’s get to know the technology behind vehicle automation in the following topic.

Currently, there are no legally operational and fully autonomous vehicles in the world. However, partially autonomous vehicles, such as cars and trucks with varying amounts of automation, from assistance for braking to aid changing lanes and parking, with some models even having a certain degree of automatic steering.

Although it is still in its infancy, autonomous driving technology is becoming increasingly common and could radically transform our transport system.

Not every Autonomous vehicle is made equal: the 6 levels of autonomy.

Different cars are capable of varying levels of autonomy, described on a scale of 0 to 5, and essential to an understanding before we talk about an autonomous vehicle’s operation.

The more technological solutions in actuators and sensors the automobile incorporates, the greater its degree of automation. As there are several stages in development, regulations and technical definitions also need to adapt.

For this reason, the Society of Automobile Engineers (SAE) created a classification to differentiate vehicles according to their degree of automation, making it easier for consumers and maintenance professionals to identify the models. The following five levels have been determined:

• Level 0: humans control all significant systems
• Level 1: specific systems, such as cruise control or automatic braking, can be controlled by the vehicle, one at a time.
• Level 2: the vehicle offers at least two simultaneous automatic functions, such as acceleration and steering, but requires human beings for safe operation
• Level 3: the vehicle can manage all critical safety functions under certain conditions, but the driver must take over when alerted
• Level 4: the vehicle is fully autonomous in some driving scenarios, although not all.
• Level 5: the vehicle is fully capable of autonomy in all situations

As we can see, while some models are already produced in series with some level of automation, other prototypes and projects are being touched by automakers and technology companies.

So it is interesting that mechanics and other maintenance professionals start to prepare for the near future. In the next topic, we’ll talk about it.

With different degrees of autonomy, autonomous vehicles become popular in the coming years and help make our daily lives much more comfortable.

Based on automakers and technology estimates, level 4 autonomous cars may already be sold in the 2–3 coming years.

How do autonomous vehicles work in practice?

The development of autonomous vehicles is at an advanced stage. Artificial Intelligence today, using computer vision and other methods, allows the vehicles to differentiate the types of obstacles and situations on the roads so as not only to react according to pre-established parameters but also to learn eventualities.

With connectivity, on-board computers exchange information with each other. Thus, an unforeseen occurrence with one of the automobiles will help everyone learn to deal with identical circumstances. For this, automobiles will need more powerful computerized central than current electronic modules (ECUs) and OBD (On-Board Diagnostic) diagnostic systems.

Most self-driving systems create and maintain an internal map of their surroundings based on information obtained from a wide range of sensors, such as radar. Some autonomous vehicles use laser beams, along with other sensors, to build the internal map. Others use radar, high-powered cameras, and sonar, and maps loaded on their systems for operation.

The software then processes the information obtained in real-time, traces a path, and issues instructions to the vehicle’s actuators that control acceleration, braking, and steering. Rules, algorithms to avoid obstacles, predictive modeling, and discrimination between objects (that is, knowing the difference between a bicycle and a motorcycle) help the software follow traffic rules and navigate obstacles.

While partly autonomous vehicles (levels 0,1,2, and 4) may require a human driver to intervene if the system encounters conflicts, the future fully autonomous vehicles (Level 5) may not even offer a steering wheel.

Autonomous vehicles can be further distinguished as connected or not, indicating whether they can communicate with other vehicles and with the city’s infrastructure, such as the next generation of traffic lights and traffic management in cities.

The technology behind autonomous cars?

The idea for autonomous vehicles is to reach the point where the driver is not required, allowing all occupants to enjoy the trip. For this, technology needs to evolve to a level of reliable security.

Several innovations are being developed, improved, and integrated into autonomous cars to make this scenario possible. Let me share here some of the most important ones:

The Hardware

They are responsible for detecting environmental characteristics and passing this data to the on-board computer. Currently, the most used are cameras, radars, sonars, and LIDARs.

Stereoscopic camera

Also called a stereo camera, it is a device that uses two or more lenses to create frames from different perspectives. In this way, you can get a sense of depth (3D), simulating human vision.

Infrared camera

The infrared camera allows accurate viewing in low or no-light environments. This equipment can identify objects by the temperature variation using sensors, capturing their infrared radiation invisible to the naked eye.

Radar

A radar emits radio waves in a specific direction, which reverberates through obstacles. By measuring the speed and intensity of this return, you can have notions of size and distance.

Sonar

Sonar works similarly to radar. Instead of radio waves, the difference is that it uses sound waves, inaudible to the human ear.

LIDAR

LIDAR also follows the logic of the two previous devices. However, laser pulses, which form thousands of luminous points, are used to scan the environment. In addition to having a faster signal, LIDAR allows you to cover a wider area, 360 °, and greater precision.

ESC (Electronic Stability Control)

Electronic Stability Control is the same technology used in several models, including in Brazil. It is responsible for calculating and making corrections in driving according to each wheel’s speed, inclination, and yaw of the vehicle in autonomous cars.

iBooster

The vacuum electromechanical servo brake, called iBooster, can generate controlled pressure on the brakes in less than 120 milliseconds. This is three times faster than conventional brake systems, making the vehicle safer in emergency braking.

GPS, speedometer, and odometer

For the vehicle to guide itself through the cities, it is necessary to equip it with updated maps and control its location. Therefore, it will use GPS equipment integrated with the speedometer and odometer. Thus, the computer can calculate its position even in the absence of the satellite.

The software

While the autonomous car’s hardware components allow the vehicle to perform such functions as seeing, talking and moving, the software is like the brain that processes environment information to know what action to take-whether to drive, stop, slow, etc. Three systems can categorize autonomous vehicle software: perception, planning, and control.

Perception

The perception system refers to the autonomous vehicle’s ability to recognize what necessary information enters through sensors or V2V components. It helps the car to understand from a given frame whether an entity is another vehicle, a person, or something else entirely.

This mechanism is similar to how our brains translate knowledge into meaning by sight. Our eye photoreceptors (sensors) detect light waves from the atmosphere and transform light waves into electrochemical signals. Neuron networks transfer these electrochemical signals back to the brain’s visual cortex, where our brain understands what these electrochemical signals Thus, our.

Thus, our brain understands whether a specific light pattern hitting our retina represents a chair, plant, or another individual.

Planning

The planning system refers to the autonomous vehicle’s ability to make certain decisions to achieve higher-order goals. And the autonomous vehicle knows what to do in a situation — whether to pause, move, slow down, etc.

The planning system works by integrating collected information about the environment (i.e., from sensors and V2X components) with defined policies and expertise about how to move in the environment ( e.g., do not drive over pedestrians, slow down when approaching a stop sign, etc.) so that the car can decide what action to take (e.g., overtake another vehicle, how to reach the destination, etc.).

As the autonomous car planning system, the human brain’s frontal processes lobe processes think and make decisions like what to wear in the morning or what to do for fun on the weekend.

Control

The control system concerns the process of translating the planning system’s objectives and priorities into actions. Here, the control device informs the necessary inputs' he necessary inputs' hardware (actuator leading to the desired motions.

For example, an autonomous vehicle, realizing it should slow when entering a red light, converts this awareness into braking practice.

In humans, the cerebellum processes play an analogous function. The cerebellum is responsible for the essential motor control function.

It encourages us to chew when the desired goal is to eat.

Artificial Intelligence and Connectivity

Artificial Intelligence would be responsible for collecting all sensor signals, internal and external, monitoring driving, notifying the owner of maintenance needs, making minor device changes and improvements, and learning from failure.

Connectivity with other autonomous cars can exchange experiences and solutions.

Such technologies make vehicle automation possible.

Vehicles that can see the world.

It’s easy to fall into the illusion that autonomous vehicles perceive their world in the same manner as we do — by “seeing” the environment using stereoscopic vision to assess the vehicle's relative locations and its surrounding elements, but this is not the case.

An Autonomous vehicle's vision system often consists of several high-resolution cameras covering the vehicle’s front, sides, and back. Unlike lidar, cameras can detect color, which is useful for spotting traffic lights, construction zone signs, and emergency vehicle lights.

These cameras are designed to work in daylight and low-light situations, but like any other camera, resolution drops as the light decreases.

When the car starts its journey, the software interprets hundreds of objects in different ways, such as cyclists or pedestrians, and reads traffic lights and signs.

Do you foresee the unusual

What if an object falls from a truck in front of you? What if a cyclist hidden between two cars wins the street as if he came out of nowhere? Google says it has been training the autonomous vehicle to handle unexpected situations. In such cases, the car’s decision is generally to slow down and wait until more information is captured about what is around.

The Autonomous driving revolution

The expectation is that autonomous Vehicles level of 4 or 5 (full autonomy) become a primary global market in 2030, moving somewhere around $ 60 billion, according to statistics data platform Statista.

By 2035, North America should have 29% of the world’s autonomous fleet, with China and Western Europe having 24% and 20% of the autonomous car fleet.

The technology of autonomous vehicles is very complicated. In the UK, 73% of all cars are expected to have some level of range (levels 1 to 3) before fully autonomous vehicles start to enter the market, predicted in 2025.

One reason behind this is the lack of consistent high-speed internet connection to allow self-driving vehicles to communicate and gather information about driving conditions and congestion or possible obstacles that block the road.

Another reason is that some vehicles require extremely detailed maps to navigate safely.

Automakers and technology companies are investing heavily in the autonomous vehicle market. No wonder. Research indicates that 85% of adults in the United States would feel safe sitting behind the wheel of a driverless vehicle.

Conclusion

This new wave of vehicles will have a monumental impact on our lives. It is already redesigning the transportation chain. This is because the autonomous (and the “semi”) do not only need exhaust, engine, and gear to run. That’s the least of it.

They embed “under the hood” a myriad of electronic devices, such as sensors, chips, radars, connectivity points, cameras, and, to sum up, all sorts of algorithms.

In other words, vehicles are becoming computers with tires.

All of this requires that automakers have new suppliers — or that old ones shake off. Hence the origin of all change. There is also a lot of money involved in this transformation. I will write about it in another article.

Autonomous Vehicles have long lived in our imagination, and if we can imagine, we can do it.

Since the Jetsons, insinuations in the future vehicles would be smart enough to take us wherever we wanted, even more so by air and not stuck to the roads.

However, the great challenge of mastering gravity has not yet been achieved, but we continue to try on the roads.. and the times seem to be mature now.

If you want to know more about how AV will change our lives and our society, please read the next article:

How Autonomous Vehicles will redefine the concept of mobility.

Autonomous cars are already among us, and some actions are already taking regarding auto repair shops and dealer…

One more thing…

If you want to go further on your learning journey, I’ve prepared for you a great list with more than 60 training courses about AI, Machine Learning, Deep Learning, and Data Science that you can do right now for free:

• The best free courses to learn AI, ML, and Data Science today.
• More than 60 courses with ratings and a summary (Made by AI, of course).
• Is it the end of the work as we know it?
• A brief analysis of the report Future of Job 2020 by the World Economic Forum

This article was originally published on Medium.com