Introduction to Automated Vehicles (CDIO #Assignment 1)

With fast development of automated technologies and computational ability, the aunomous driving of vehicles has become one of the hottest topics among both coventional autonomous companies (eg. GM and Ford) as well as IT companies (eg. Googleand Baidu).

The first autonomous driving vehicle system is called "NavLab" from Carnegie Mellon University in USA in 1980s. After the autonmous vehicle competition called "Urban Challenge" in Amercian soon become famous and spread the fast development of autonomous technologies all around the world. 

From modifying coventional vehicles to design autonomous systems from beginning of vehicle development, the autonomous driving technologies have seen many impressing breakthroughs over the 30 years. And it is even changing faster these five years than ever.

Within this post we will discuss some basic ideas about autonomous driving, like the legal classification of AD (autonomous driving) levels and analysis some examples. Later we can discuss more about the system architecture from vehicle design point of view and also the challenges of AD now companies are facing. 

So now let's see the basic AD classficaiton levels and some examples.

1.SAE classification of autonomous driving levels

There in total 6 level of AD defined by SAE based on how much human driver take control of the vehicle, or on the contrary, how much the vehicle could drive itself by controlling from systems.Here the "system" refers to the driving assistance system, or automated driving system. This classification is rather technological than legal.

The contents and description of each level can be seen from Table 1 below.

Table 1 SAE classification of AD level

As seen above, from level 0 (No Automation) to level 5 (Full Automation), the different levels are mostly classified by how much human drivers or system take the responsibility in driving. 

From automotive engineers' point of view, a vehicle system contains three parts: human driver, environment and the vehicle. These three parts are connected and combined together to function a proper working driving task.

As the core of vehicle controlling, the driver model controls mainly two input to vehicle system: steering and pedals  (Acceleration and braking). And the analysis of following six levels also focus on controlling of these inputs. Of course, for automated vehicle, the vehicle intelligent system could take over some part of all of the responsibilities from human drivers.

Level 0: This is the basic conventional driving mode of cars. The driver (human) controls everything: steering, brakes, and power. All the performance are depend on drivers without active assistance from vehicle system.

Level 1: In this level most functions are still controlled by the driver, but a specific function can be controlled automatically by the vehicle. This means one of the input (steering or acceleration/braking) can be automated. 

Level 2: In level 2, at least one driver assistance system controls both "steering and acceleration/ deceleration using information about the driving environment" is automated, like adaptive cruise control and lane-centering. It means that the "driver is disengaged from physically operating the vehicle by having his or her hands off the steering wheel AND foot off pedal at the same time," according to the SAE. However, the driver must still always be ready to take control of the vehicle. Most ADAS (Advanced driving assistance systems) that already on the market are between level 1 and level 2.

Level 3: Human drivers are still mostly in control of vehicle until it comes to level 3. In this level the car is able to monitor the environment, detect and analysis environment information automatically. Under certain conditions drivers can just let vehicle take control without providing environment information. However, drivers are still necessary in level 3 cars,  which means that the driver is still present and will intervene if necessary, but is not required to monitor the situation in the same way it does for the previous levels. 

Level 4: In this level the vehicle has reached the fully autonomous mode, and are able to perform all safety-critical driving functions and monitor roadway conditions for an entire trip. However, it's important to note that this is limited to the "operational design domain (ODD)" of the vehicle—meaning it does not cover every driving scenario.

Level 5: This highest level indicates a fully-autonomous system that expects AD system can perform just the same as that of a human driver, in every driving scenario—including extreme environments like dirt roads that are unlikely to be navigated by driverless vehicles in the near future

2. Examples of Autonomous cars

The leading positions of autonomous vehicle research companies have been changing rapidly recent years. There is a diagram showing the overall potential of top car companies about autonomous driving in 2017.

Figure 2.1 Leaderboard of AD research companies

Although the results can be further discussed from different points of view, it indicates a fact that nowadays autonomous driving is no longer just the job for car companies. It's more about AI, IT technologies and also data base!

So let's take two typical examples: one from conventional car companies (Volvo) and one from IT companies (Google).

(1) Volvo Car

Safety is always the most concerned factor for Volvo, as it's famous for "the safest car in the world". Volvo Cars put great effort in both passive and active safety. And when active safety comes into autonomous driving, the safety factor is definitely the most significant point to be considered.

Volvo car has made great progress on autonomous driving since 2006. the following table shows the ADAS and AD systems they launched each year.

Year	ADAS Functions	Introduction
2006	Adaptive cruise control (ACC)	Allows car to maintain a set distance from front car
	Lane departure warning (LDW)	Alerts driver and activates when beginning to stray across lane markings
2007	Collision warning with auto brake	Assists driver by detecting an imminent rear end collision then apply brake if necessary
2008	City safety	Avoid front colliding at low speed by using LIDAR and apply brake automatically
2009	SARTRE (Start of Safe Road Trains for The Environment)	A pioneering project to develop platooning of AD cars on highway
2010	ACC with queue assist	Enhance of ACC that works in slow-moving traffic and also when traffic comes to a halt
	Collision warning with full auto brake	Radar and camera support to detect rear end collision and applies full braking
	Pedestrian detection with full auto brake	Vision based pedestrian detection and auto brake
2012	Traffic sign information	Recognizes road signs and displays them on dash.
	Park Assist Pilot	Helps parking by identifying a suitable space and steering the car into it.
	Lane-keeping aid	Assisting driver by steering car to keep it with in lane
	Active high-beam control	Conveniently dips main beam automatically when approaching oncoming vehicles
	SARTRE completed
2013	Autonomous parking	Self-parking controlled by mobile phone
	Drive Me project announced	World’s first large-scale trial of self-driving cars
2014	Pilot Assist	Combines lane keeping and ACC
	Auto brake at intersection	Protects drivers from oncoming cars by auto-braking
2017	Drive Me pilot	AD on real roads

Table 2 Volvo Car ADAS and AD development

Source: http://www.volvocars.com/intl/about/our-innovation-brands/intellisafe/autonomous-driving

From the above development process, we can see that Volvo car is progressively completing its autonomous driving project by combining ADAS functions into the whole system. While for active safety functions and ADAS systems, Volvo usually test it in the test track called ASTAZero. And for the urban road testing project "Drive Me", it will take place on public roads, and will involve local drivers integrating autonomous driving technology into their daily lives. 

Figure 2.2 Volvo XC90 Drive Me Project

The goal of Volvo cars in year 2020 is that no one will be fatally hurt within a Volvo. We can see that Safety is still most concerned when coming to autonomous driving.

(2) Google

Google has start the self-driving car project since 2009 when it starts the company called Alphabet. Now the project is taken over by a new company called Waymo, which is an autonomous car development company spun out of Google's parent company, Alphabet Inc., in December 2016. 

Figure 2.3 Waymo car

The project team has equipped a number of different types of cars with the self-driving equipment, including the Toyota Prius, Audi TT, and Lexus RX450h, Google has also developed their own custom vehicle, which is assembled by Roush Enterprises and uses equipment from Bosch, ZF Lenksysteme, LG, and Continental. In May 2016, Google and Fiat Chrysler Automobiles announced an order of 100 Pacifica hybrid minivans to test the technology on.

From the data we get to know that Google robotic cars have about $150,000 in equipment including a $70,000 LIDAR system.The range finder mounted on the top is a Velodyne 64-beam laser. This laser allows the vehicle to generate a detailed 3D map of its environment. With this high accuracy information of environment, Google self-driving car can easily build the map of surrounding and carried out the SLAM (Simultaneously Localizaiton and Mapping).

The following table shows the progress of Google's self-driving car project.

Year	ADAS Functions	Introduction
2009	Self-driving car project began	Succeeded in driving an order of magnitude larger than had ever been driven autonomously in Toyota Prius vehicles.
2012	More than 300000 miles of self-driven	Early testing on highways
	Move to complex city streets	Focus on more complex environment of city streets.
2015	“Firefly” hit public roads for the first time	Fully self-driving cars named “Firefly” hit the road.
	World’s first and only fully self-driving ride on public roads	No steering wheel, no pedals and no driver
2016	Waymo, a self-driving technology company
2017	Introduced fully self-driving Chrysler Pacifica Hybrid minivans	first vehicle built on a mass-production platform with a fully-integrated hardware suite, newly designed by Waymo for the purpose of full autonomy.
	Launched early rider program	Invited residents in Phoenix, AZ to join a public trial of our self-driving vehicles and help shape the future of how our cars will work.

Source: https://waymo.com/journey/

Compared with conventional car companies like Volvo, we can see that IT companies seen to develop autonomous technologies from a different point of view, and even faster. 

Rather than worried about massive production and apply ADAS functions to on-road vehicles, Google focus more on the environment sensing and mapping algorithm. Data seem to be the most important point for higher level of autonomous driving technology and algorithm. 

But similar with Volvo, after launching autonomous vehicle with certain level, they both start the real road test project in order to further analysis the complex using conditions and gain deep understanding and modification according to real use cases.

3. Analysis of above two examples

(1) Volvo cars

Volvo cars has promoted many ADAS functions. Since it has not reached fully autonomous driving mode, we can take the on-market function Pilot Assist for an example for further analysis.

Luck enough as an automotive student at Chalmers, I had the chance to experience this Pilot Assist function few weeks ago at ASTAZero at a highway road. The experience was amazing. 

I would like to classifiy the volvo activated with Pilot Assist function into Level 2.5 for autonomous driving. This should be somewhere between level 2 and level 3. Because the Pilot Assist system is a combination of lane keeping, adaptive cruise control and some other active safety systems. The driver can be free from both steering wheel and padels control. This make it come into Level 2. And the vehicle system is getting environment information through cameras, GPS and other sensors to control the vehicle and even try to understand the driving intention of drivers. However, the driver cannot be free all the time from monitor the enivronment. After ten second without touching steering wheel, the vehicle would give some warning to keep the driver pay attention to the condition. So we can say this system is not fully fulfill the Level 3 requirement, but somehow above Level 2.

(2) Google

Since companies like Waymo does not need to care much about the vehicle overall design and performance, like vehicle dynamic performance, these companies can make some breakthrough out of conventional cars' boundary.

From the information shown on website of Waymo, we can see that Google's car has already achieved as least Level 4 of autonomous driving, which can automatically sense the environment and drive the car without any attention from human driver. But since it's still in test of various real road conditions, we cannot come to conlusion that Google's car can overcome any driving conditions and reaches the Level 5 of autonomous driving.

4. Applications for autonomous vehicles

(1) Public transportation with higher safety and efficient level

With high level of autonomous public transportation vehicles like bus and tram, there will be highly efficient traffic condition and no more traffic jam. The traffic can be easily planned and controlled. People can save much time on their way home or to work. Also, without human driver, the public transportation system can be more safe with less rish of accidents.

(2) Health care

Special designed autonomous vehicles can be used for elderly health care. Old people who are not convenient to walk or take public transportations can use this kind of vehicle to transport safely and easily. Also the intelligent systems on the vehicle can detect the situation of human and provide service and help if necessary. It would be just like an intelligent ambulance.

(3) Replace human on dangerous working conditions

Imaging the hard working condition that human would not be able to handle, like deep sea, or out of earth, maybe on Mars. Autonomous vehicles could have the benifit to replace human and take care of these work. For example, the autonomous vehicles used to explore the surfeace of Moon or Mars. Due to long distance from earth, the vehicle cannot simply be remote controlled. Therefore, autonomous vehicles have to ba able to sense the environment and carry on some works on its own. With this kind of vehicle or robot, dangerous work will no longer need human to rish their lives. To some extends, this can also save large amount of money.