Autonomous Car

2.3.1 Introduction of the Autonomous Car

The autonomous car, which is also called a self-driving, driverless or robotic vehicle, referred to a machine car using multiple car sensors, such as infrared radar and GPS, to perceive the surrounding driving conditions and send data to the master computer to autonomously make decisions, such as adjusting the speed or direction. In other words, the autonomous car was a car driven by a machine.

In 2016, the National Highway Traffic Safety Administration (NHTSA) adopted the driving automation levels defined by SAE International (Warrendale, 2016). The six levels and their definitions are shown in Table 5.

Table 5. SAE J3016™ Levels and Definitions

Level Level title Definition

0 No Automation

The full-time performance by the human driver of all aspects of the dynamic driving task, even when enhanced by warning or intervention systems.

1 Driver Assistance

The driving mode-specific execution by a driver assistance system of either steering or acceleration/deceleration using information about the driving environment and with the expectation that the human driver performs all remaining aspects of the dynamic driving task.

2 Partial Automation

The driving mode-specific execution by one or more driver assistance systems of both steering and acceleration/deceleration using information about the driving environment and with the expectation that the human driver performs all remaining aspects of the dynamic driving task.

3 Conditional Automation

The driving mode-specific performance by an Automated Driving System of all aspects of the dynamic driving task with the expectation that the human driver will respond appropriately to a request to intervene.

4 High Automation

The driving mode-specific performance by an Automated Driving System of all aspects of the dynamic driving task, even if a human driver does

15

not respond appropriately to a request to intervene.

5 Full Automation

The full-time performance by an Automated Driving System of all aspects of the dynamic driving task under all roadway and environmental conditions that can be managed by a human driver.

(Reference: SAE International, 2016)

The biggest difference was encountered between Level 2 and Level 3. From Level 0 to Level 2, a human was mainly responsible for monitoring the car driving, whereas in Level 3 to Level 5, the autonomous driving system was responsible for monitoring the car. Currently, car manufacturers such as Mercedes-Benz, Audi, and Ford have launched commercial cars reaching Level 2.

2.3.2 Information Technology of the Autonomous Car

The overall system of the autonomous car could be divided into the following three modules: (1) Perception module: perceived and collected the environment data and position data. (2) Decision module: analyzed the data and made decisions on how to drive the car. (3) Control module: controlled the car with the steering, accelerator, brake and gearbox according to the decisions from the computer (Tettamanti, T., Varga, I., & Szalay, Z., 2016; Paden, B., Čáp, M., Yong, S. Z., Yershov, D., & Frazzoli, E., 2016). The modules and process are shown in Figure 4.

Perception module was similar to the eyes and ears of the autonomous car, responsible for perceiving and collecting the environment data and position data. The IT of the perception module included environment perception and positioning technology. Environment perception technology was used to identify the surrounding conditions and collect data as the basis for decision-making. It collected the local car data with the embedded sensors and cameras and global environment data from the cloud, V2V (Vehicle-to-Vehicle) and V2I (Vehicle-to-Infrastructure) technology.

Position technology including GPS and mapping service was used to determine the car’s location, which was the critical basis for route planning and mission planning. It acquired the map data and the real-time position of the car (Narla, 2013; Lutin, Kornhauser, & Masce, 2013).

Decision module was the brain of the autonomous car, responsible for analyzing data and making decisions on how to drive the car. There were two stages in the autonomous car’s decision-making, mission planning and behavior decision. Mission planning technology performed the global thinking on how to complete a mission such as going to the destination and parking the car. Route planning technology was the basis for implementing automatic driving, which found a non-collision route in the

16

environment with obstacles from initial status to target status. Behavior decision, another stage of decision, was responsible for deciding the next action based on the data collected by perception technology, and for controlling the car

Control module, also known as the hands and feet of the autonomous car, was responsible for controlling the steering and accelerator according to the decision from the decision module.

Figure 4. The Module and Decision Process of the Autonomous Car

2.3.3 Decision-making of the Autonomous Car The Data and Environment of Autonomous Cars

Driving on the road was a constantly moving process, and the autonomous car had to make ten decisions per second, deciding whether it should change direction or adjust the speed (Google, 2015). Therefore, the system environment of autonomous driving was dynamic (rapid-changing), unpredictable and open so that during processing the system must consider the external factors or interact with external objects. The decision was unstructured and full of uncertainty.

The key which drove autonomous car being progressive included the sensor/detector, Internet and AI technologies. The advancement of sensor/detector and Internet technology significantly increased the range and accuracy of data collection of the perception module, whose quality greatly affected the information and decision quality (Laudon & Laudon, 2012); meanwhile, the advancement of AI technology promoted the velocity and accuracy of decision-making od the decision module and made the car become smarter. (Fu, 2016)

The detectors of the autonomous car sampled data points thousands of times per second to acquire the most real-time data (Lutin, Kornhauser, & Masce, 2013).

Essential data included position and environmental data from the devices of the car and the cloud, which we listed in Table 6.

The main data resources of autonomous car included sensors/detectors and

17

V2V/V2I via Internet. With the current technology, because it was difficult to make all cars be equipped with V2V/V2I technology, the cars could not communicate with each other and could only mainly depend on the sensors/detectors. To increase the accuracy of environment perception, autonomous car was embedded with multiple types of sensors/detectors, including LIDAR, Radar and camera technology.

LIDAR, the abbreviation for light detection and ranging, was the master of 3D mapping, and applied to the rotatable detector installed on the roof of the car to measure the distance to the target or construct the 3D image by injecting a pulsed laser light and measuring the reflected pulses to determine the current location and situation of the car.

Similar to the principle of LIDAR, Radar, the abbreviation for radio detection and ranging, was responsible for the motion measurement to determine the range, angle, or velocity of objects, and applied to the sensors installed in the car to understand the surrounding and construct the road situation around the car by injecting a radio wave.

Compared to the above two technologies, camera was the cheapest detector of the car.

Cameras could collect the image and can see the color. (Santo, 2016; Wang, 2016) From above, we found that it was greatly different from the past, the data type was not only the text or the digital type, but also included the pixel, image, 3D mapping, and so on.

The data type had changed – from the single dimension into multiple dimension, and the data process and analysis also became more complicated. In addition, the data size was larger. Because of the diverse data variety and volume, the data and decisions of the autonomous car were exactly a type of big data that was high-volume, high-velocity, high-variety and high-veracity.

Table 6. The Data of the Autonomous Car

Type Data source Data

Position data

GPS Position

Mapping service Global map

Environmental

V2V Other cars’ behaviors

V2I Traffic conditions, such as parking lot, toll, road condition

The Decision Circumstances of Car Driving

The following table shows the common driving scenarios and decision

18

circumstances. The driving scenarios were divided into three categories including route planning, general driving and parking/reversing. There were three elements of driving decision that drivers frequently needed to make to control the car. The three elements were route, direction and speed. The column on the far right of the table showed how people make driving decisions without IT assistance.

Table 7. The Common Driving Decision Circumstances

Driving scenario Route Directio

n Speed Driving without IT Route Planning Search the location of destination v Look for the location by map

Decide the route v Decide the route by map

General Driving

Control speed v Control the gas pedal for acceleration or

the brake for deceleration

Control direction v Control the steering wheel to adjust the

direction Dodge obstacles

(Emergency) v v Judge the appropriate distance to obstacles

to adjust the direction or speed Maintain safe driving distance

(Normalcy) v v Judge the appropriate distance to other cars

to adjust the direction or speed Continue driving in the right lane v Pay attention to the instantaneous road

conditions to control the driving direction

Parking/

Reversing

Search the parking place v Look for a parking place while driving

Decide the parking/reversing route v Decide the available parking/reversing route by review/side mirror and observing

Parking/reversing v v Control and adjust the speed and direction

to match the route