The Levels of Autonomous Driving: Dedicated Short-Range Communications (DSRC)
The Levels of Autonomous Driving: Dedicated Short-Range Communications (DSRC)
Technology
Autonomous driving has the potential to improve the safety and efficiency of the transportation system.
Sensors are the key components of autonomous vehicles.
tasks.
The System
controls either the
L1 Hands or feet off. vehicle’s speed or
steering. Ex: any car with auto braking or
cameras and RADARs lane assist
Ex: none
GPS IMU
Inertial Measurement Unit
Encoders
- used for control and - used to provide
provides
guidance of odometer data on
geolocation and
autonomous vehicles the vehicle
time information
-provides information
- measures a body's on position, direction
force, angular rate and velocity within a
and magnetic field control system 3
Connected Vehicles (CV)
Connected
Vehicles (CV)
100Mbit/s
15Mbit/s
<0.1Mbit/s
Camera Ultrasonic RADAR LIDAR Vehicle motion,
GNSS, IMU
SOURCES
[1] Lin, Shih-Chieh. Zhang, Yunqi. Hsu, Chang-Hong. Haque, Matt Skach Md E. Tang, Lingjia. Mars, Jason. The Architectural Implications of Autonomous Driving: Constraints and
Acceleration. ASPLOS '18 Proceedings of the Twenty-Third International Conference on Architectural Support for Programming Languages and Operating Systems Pages 751-
766, Williamsburg, VA, USA, March 24 - 28, 2018. CREATED BY
[2] Rudolph, Gert, and Uwe Voelzke. “Three Sensor Types Drive Autonomous Vehicles.” FierceElectronics, 10 Nov. 2017, www.fierceelectronics.com/components/three-sensor-
types-drive-autonomous-vehicles.
[3] Campbell, Sean. Mahony, Niall O’. Krpalcova, Lenka. Riordan, Daniel. Walsh, Joseph. Murphy, Aidan. Ryan, Conor. Sensor Technology in Autonomous Vehicles : A review.
2018 29th Irish Signals and Systems Conference (ISSC), Belfast, UK, 21-22 June 2018. IEEE
[4] SCHOETTLE, BRANDON. SENSOR FUSION: A COMPARISON OF SENSING CAPABILITIES OF HUMAN DRIVERS AND HIGHLY AUTOMATED VEHICLES. SWT-2017-12. AUGUST 2017
[5] Heinrich, Stephan. Flash Memory in the emerging age of autonomy, Flash Memory Summit 2017, Santa Clara, CA, Flash Memory Summit. Lucid Motors.
Tergel Molom-Ochir
[6] [email protected]. “Automated Vehicles for Safety.” NHTSA, 12 Aug. 2019, www.nhtsa.gov/technology-innovation/automated-vehicles-safety.
UMass Amherst
Dec, 2019
Autonomous Driving and Its Sensor
Technology(Supporting Document)
Tergel Molom-Ochir
University of Massachusetts
Amherst
Amherst, Massachusetts
Abstract— This supporting document will address at this level. Any car with auto-braking or lane assist
and clarify topics from the infographic “Autonomous feature is level 1 [1][4].
Driving and Its Sensor Technology” by reviewing the • Level 2 – Partial Automation
levels of autonomous driving and the sensor packages
The system can control both the vehicle’s speed
used to create autonomous vehicles. For any type of
autonomous vehicle, sensors are crucial to sense the and steering. Level two autonomous vehicles have
environment to make calculated decisions. This Automatic Emergency Braking (AEB) plus Lane
supporting document elaborates on the level of Keeping Assist (LKA) features. The human driver
autonomous driving, sensors, their advantages and can be hands and feet off. At this level, the human
disadvantages, and a comparison between the sensors. driver still monitors the environment and driving.
Cameras and RADARs are widely used at this level.
Keywords: levels, sensors, autonomous driving. Vehicle control is shared between the human driver
and the system. Companies such as Tesla and
I. INTRODUCTION Mobileye are known for producing Level two
Driving is a necessity for today’s life. In recent autonomous vehicles [1][4].
years, research in the field of autonomous vehicles • Level 3 – Conditional Automation
has advanced remarkably due to the research and The system can perform driving tasks under
development in the field of sensors [3]. As a limited driving conditions, but the human driver will
consequence, the levels of autonomous driving have be needed to intervene and resume driving at times.
been defined and the advantages and disadvantages The human driver engagement can be considered to
of each sensor have been evaluated [3]. In this be "Eyes-Off engagement." Starting from this level,
supporting document, we review the levels of the system monitors the environment and driving.
autonomous driving (Section 2) and the sensors that Cameras, RADARs, and LIDARs are commonly
are frequently used in modern autonomous vehicles, used at this level. Vehicle control is shared between
their advantages and disadvantages, and a the human driver and the system. Companies such as
comparison between the sensors (Section 3) expand Nvidia/Audi and Waymo are known for building
upon and clarify topics from the infographic experimental level three autonomous vehicles [1][4].
“Autonomous Driving and Its Sensor Technology.” • Level 4 – high automation
II. LEVELS The system will perform all driving tasks even if
To better inform the public, the National the human driver does not respond. Driver
Highway Traffic Safety Administration of the United engagement can be considered to be “Brain-off
States Department of Transportation published engagement”. The system monitors the environment
“Automated Vehicles Safety” in which they and driving. The automated system has full control
discussed timeline, levels of autonomous driving and over the vehicle. There are no examples because level
benefits of automation [6]. The following are the four automation has not been developed yet [1][4].
levels of autonomous driving. • Level 5 – Full Automation
• Level 0 – No automation The system will perform all driving tasks
This level has no automation. The human driver without any help from the human driver. The human
completes all driving tasks under any driving driver does not need to be present. Driver
conditions. engagement can be considered to be “Brain-off
• Level 1 – Driver Assistance engagement”. The automated system has full control
The system can control either the vehicle’s speed over the vehicle and monitors the environment and
or steering. The driver can be hands or feet off. At driving. Level five has not been developed yet;
this level, the human driver monitors the environment therefore, there are no examples [1][4].
and driving. Cameras and RADARs are widely used III. SENSORS
Figure 1 presents two main types of sensors: systems (advanced driver-assistance systems) are
exteroceptive sensors and proprioceptive sensors. common in modern vehicles [4]
Exteroceptive sensors are used to get information 4) Ultrasonic
about the surrounding environment while An ultrasonic sensor uses sound waves to
proprioceptive sensors are used for measuring values measure the distance to an object by emitting it
within the system [3]. As shown in Figure 1, camera, towards the object and measuring the time it takes for
LIDAR, RADAR and ultrasonic sensors are the waves to return. The main advantage is that they
exteroceptive sensors, used to perceive the are the cheapest exteroceptive sensor that can be
environment while GPS, IMU, and encoders are found for autonomous vehicles. Ultrasonic sensors
proprioceptive sensors, used to provide motor speed, are good at short range measurements; they are the
wheel position, latitude and longitude, etc. most accurate sensor for close proximity applications.
A. Exteroceptive sensors – Autonomous Vehicles However, the main disadvantages of these sensors
1) Camera come from disturbances in the sound waves. The
Because with cameras, color vision is sound waves can be disturbed by temperature and
possible, autonomous vehicles will be able to identify humidity as sound travels in a medium [3].
road signs, traffic lights, lane markings, which is
crucial for dealing with traffic. A real-time video B. Proprioceptive sensors – Autonomous Vehicles
object detection is used for autonomous vehicles. The 1) GPS
field of view is ~ 45° to ~ 90° for cameras. GPS is a system that provides geolocation
Furthermore, the range varies for different types of and time information by connecting the GPS receiver
cameras with different focal lengths but generally, to GPS satellites. For autonomous vehicles and
the range is ~150 meters for monocular systems and driving, GPS sensors are crucial for navigation and
~100 meters for stereo systems. Resolutions vary by localization. The main advantage is that they are
a large margin [3]. The main advantage cameras have accurate; their accuracies can range from a few
over other sensors is that they can see colors and centimeters to 3 meters. The main disadvantage,
textures. In contrast, the main disadvantage of however, is that because GPS sensors need a direct
cameras is that they can be affected by weather line of sight with the satellites, navigation indoor or
conditions and they are sensitive to low-intensity underground cannot be calculated [3].
lights [4]. 2) IMU
2) LIDAR IMU, an Inertial Measurement Unit, is a
LIDAR, which stands for Light Detection device that measures force, angular rate, and
and Ranging, is a remote sensing technology that is magnetic field of a body using three accelerometers,
used for determining accurate distance and size gyroscopes, and magnetometers. From the raw IMU
information of objects in the surrounding data, linear velocity, altitude, and angular positions
environment. LIDAR operates by sending out a (relative to a global reference frame) of autonomous
pulsed laser of light and measuring the time it takes vehicles can be calculated. There are a few
for it to be reflected. This principle is known as the disadvantages of IMU due to how they work. The
time of flight (TOF). Measurements are gathered and main disadvantage is that they only provide motion
are used to generate a 3D map of the surrounding information and not location information. Another
environment. The field of view for a LIDAR sensor disadvantage is that the data can be affected by drift
is 360°. The range is about 200m on average. It can errors. These problems can be solved by
be over 250m for long-range sensors [3]. The main implementing GPS devices [3].
advantage of LIDAR sensors is their high precision 3) Encoders
and accuracy. However, the main disadvantage is Encoders are electro-mechanical devices
their high cost and low availability [4]. that provide information on position, direction, and
3) RADAR velocity within a control system. They are commonly
Radar is a system that sends out pulses of used to provide odometer data. The main advantage
electromagnetic waves to determine the presence, is that they are cost-efficient and easy to install.
direction, distance, angle, and velocity of objects. However, the disadvantage is that slippage between
The field of view of a radar sensor is ~ 15° for long- the wheels and the ground can affect the data,
range radar sensors and ~ 90° short-range radar providing wrong localization and velocity
sensors. The range can be around 250 meters [3]. information [3].
Radar sensors offer many benefits for autonomous
vehicles, two of which are low cost and high
availability. Due to their high availability, ADAS
Figure 2. Comparison between exteroceptive sensors
V. CONCLUSION
This supporting document expands upon and
clarifies topics from the infographic “Autonomous
Driving and Its Sensor Technology” by elaborating
on the level of autonomous driving, sensors, their
Figure 1. Exteroceptive Sensors and Proprioceptive advantages and disadvantages, and a comparison
Sensors between the sensors. In recent years, with the
improve in the field of sensors, the field of
C. Dedicated short-range communication (DSRC) – autonomous vehicles has improved drastically.
Connected Vehicles Autonomous driving has the potential to improve
DSRC is a wireless communication method safety and transportation on roads.
for connected vehicles. Connected vehicles will be
able to communicate recent and future actions with REFERENCES
[1] Lin, Shih-Chieh. Zhang, Yunqi. Hsu, Chang-Hong. Haque,
each other. Each connected vehicle will be able to Matt Skach Md E. Tang, Lingjia. Mars, Jason. The Architectural
wirelessly send and receive information [3]. Implications of Autonomous Driving: Constraints and
Acceleration. ASPLOS '18 Proceedings of the Twenty-Third
IV. COMPARISON International Conference on Architectural Support for
Programming Languages and Operating Systems Pages 751-766,
On the infographic “Autonomous Driving and Its Williamsburg, VA, USA, March 24 - 28, 2018.
Sensor Technology,” comparisons between sensors [2] Rudolph, Gert, and Uwe Voelzke. “Three Sensor Types Drive
were made as shown in Figure 2. First, when the Autonomous Vehicles.” FierceElectronics, 10 Nov. 2017,
horizontal field of view was compared, LIDAR had www.fierceelectronics.com/components/three-sensor-types-drive-
autonomous-vehicles.
the most field of view of 360° and long and mid- [3] Campbell, Sean. Mahony, Niall O’. Krpalcova, Lenka.
range radars had the least field of view of 18°. Riordan, Daniel. Walsh, Joseph. Murphy, Aidan. Ryan, Conor.
Second, when comparing the ranges, long and mid- Sensor Technology in Autonomous Vehicles : A review. 2018 29th
range radars had the longest range and short-range Irish Signals and Systems Conference (ISSC), Belfast, UK, 21-22
June 2018. IEEE
radars had the shortest range of all the sensors. [4] SCHOETTLE, BRANDON. SENSOR FUSION: A
Lastly, how much data each sensor generates was COMPARISON OF SENSING CAPABILITIES OF HUMAN
displayed as a column chart. Cameras generate the DRIVERS AND HIGHLY AUTOMATED VEHICLES. SWT-
most data and ultrasonic sensors generate the least 2017-12. AUGUST 2017
[5] Heinrich, Stephan. Flash Memory in the emerging age of
amount of data among all the other sensors [4]. autonomy, Flash Memory Summit 2017, Santa Clara, CA, Flash
Memory Summit. Lucid Motors.
[6] [email protected]. “Automated Vehicles for
Safety.” NHTSA, 12 Aug. 2019, www.nhtsa.gov/technology-
innovation/automated-vehicles-safety.