These slides use concepts from my (Jeff Funk) course entitled analyzing hi-tech opportunities to analyze how autonomous vehicles are becoming economically feasible throug through improvements in lasers, microelectronic mechanical systems (MEMS), integrated circuits (ICs), and other components. Although the cost of the Google Car is currently about 150,000 USD, 30% annual improvements in lasers, MEMS, and ICs will make these economically feasible for a broad number of users in the next ten years. A key issue is when certain lanes, roads or even entire highway systems are restricted to automated vehicles. This would enable collision avoidance to rely more on between-vehicle communications. This would further reduce the cost of automated vehicles, stimulate diffusion, and also reduce transportation time and increase fuel efficiency.
2. Contents
1.
Introduction to Autonomous Vehicles
2.
Essential Technologies
1. GPS
2. Lidar
3. Radar
4. Video
5. Ultrasound
3.
The Future: Driverless Fleets
Vehicle-to-vehicle interaction
4.
System Integration
CPU and Broadband Requirements
5.
Entrepreneurial Opportunities
evaluation of cost and
performance parameters
3. Introduction
What are Autonomous Vehicles?
Self-driving cars with sensors able to observe and understand surroundings
Google driveless car estimated cost USD$150,000 (Year 2012)
5. Benefits of Autonomous Car
Google’s
Aspiration
Potential Annual Benefits
(US only)
4.95 million fewer accidents
90% reduction 30,000 fewer deaths
in ACCIDENTS 2 million fewer injuries
$400 billion saving in cost
90% reduction
WASTED
COMMUTING
4.8 billion fewer commuting hours
1.9 billion gallons in fuel savings
$101 billion saved in lost
productivity and fuel cost
90% reduction Reduce cost per trip-mile by 80%+
Car utilization from 5-10% to 75%+
in CARS
Better land use.
Ref: http://www.carloan4u.co.uk/infographics/the-ultimate-car-of-the-future/
Google, US NHTSA, AAA, Texas A&M Transportation Institute, Columbia University Earth Institute and Devil’s Advocate Group’s analysis
6. Autonomous Systems Comparison
Google Car
Oxford
Robot Car
Ionut
Budisteanu
Mobil Eye
C2-270
MuCAR-3
Drive-assist
system cost
USD 130,000
USD 7,500
USD 4,000
USD 749
N.A.
Level of
Autonomy
High
Medium
Medium
Low
High
Key
Components
GPS,
3D Lidar
Computation
software
2D Lidar,
Prior
knowledge
Low-end 3D
Lidar,
Webcam
supervisor
program
Mono
camera,
System-onchip
Multifocal
camera,
360º Lidar
Ref: http://www.driverless-future.com/?p=400
Ref: http://www.mobileye.com/blog/press-room/products-press-room/mobileye-announces-new-c2-series-solution-and-previewsnew-technologies/
7. Technology: GPS
GPS
Lidar
Radar
Video
Ultrasound
Space segment composed of GPS
satellites that transmit
time & position
in the form of radio
signals to the user
How GPS Works
Control segment composed of all the
ground-based facilities
that are used to
monitor and control
the satellites
User segment - consists of
the users & GPS receivers
9. Technology: GPS
Reference: GPS in Aerospace
GPS
Lidar
Radar
Video
Ultrasound
FAA data collected in early 2011
shows that the horizontal accuracy of
GPS SPS is often within ~ 1 meter
Reference: GPS in Aerospace
By hypothesis, the car accuracy in
2020 may be within < 50cm
Using Nationalwide Differential
GPS System ~ 10 to 15cm
Ref: Federal Aviation Administration
10. Technology: GPS
GPS
Lidar
Radar
Video
Conditions
Maintain same fuel price
General road conditions
Factors
GPS Cost Benefits
Ultrasound
GPS
Accuracy
Cost
GPS Cost Benefits
Fuel
Accident
consumption rate
Insurance
premium
Time
Enable
potential
cost saving
over time
11. Technology: Lidar
GPS
Lidar
Radar
Video
Ultrasound
• Non-contact technology
– Sense the physical appearance,
speed and distance of objects
through laser.
– Software transforms the data
into 3D representation of
objects in real time
• Mobile Mapping System
– GPS: geo-referencing, motion
sensor optional
– Laser: range finding, property
evaluation
– Software: post-processing, map
construction.
Lidar is uniquely suited to
,
surveys
• 1-1000m distance
• Mm to cm accuracy
Ref: Audi's sensors include radar, lidar, camera, and adaptive cruise-control (source: Audi)
13. Technology: Lidar
GPS
Lidar
Radar
Video
Ultrasound
Component Cost: Laser
Ref: http://www.industrial-lasers.com/articles/print/volume-27/issue-1/features/2011-annual-economic-review-and-forecast.html
Ref: Global Markets for Laser Systems, Components & Materials -- Focus on the Americas published by BCC Research on January 1, 2012
Ref: http://www.toptica.com/products/ultrafast_fiber_lasers/technology.html
14. Technology: Lidar
GPS
Lidar
Radar
Video
Ultrasound
Component Cost: Inertial Sensor
Ref: Technology Trends for Inertial MEMS, market & technology report – Jan 2012
Ref: IC Insights - “Market Analysis and Forecast for Optoelectronics, Sensors/Actuators, and Discretes”, May 2013
20. Technology: Radar
GPS
Lidar
Radar
Larger Bandwidth
Is required for better
1. Range accuracy
2. Contour determination
“79GHz Project” to allocate a broad band for
automotive Short Range Radar applications
Ref: Bandwidth
Requirements for Future
Safety and Comfort
Applications, Markus
Andres, Universität Ulm
Video
Ultrasound
22. Technology: Camera
GPS
Lidar
Radar
Camera Ultrasound
Humans need two eyes to see the world in binocular vision, which enables us to calculate
the object sizes, distances and perceive longitudinal motion.
In recent years (2011 to 2013), stereo video camera system has been created and
adopted by many leading car manufacturers, to give driver assistance systems the same
capacity: a 3-dimentional view of the area in front of a vehicle
Stereo Camera System
23. Technology: Camera
GPS
Lidar
Radar
Camera Ultrasound
Mercedes-Benz S-Class Luxury Sedan – Launched in 2013
One of the most important
development is the introduction on
Stereo Camera Technology
Highlights of the Stereo Camara in S-Class:
- Price: $2,800 for option package of stereo
cameras and radar;
- Opening Angle: 45°;
- 3D view of the road ahead: 50m;
- Total Range: 500m;
- Autonomous emergency braking with
pedestrian detection: up to 72km/h;
- Data usage: 300gb per Hour
- Distance / Height measurement Accuracy:
3 millimeters within 15m
28. Technology: Camera
GPS
Lidar
Radar
Camera Ultrasound
Back-illuminated CMOS image
sensor, which significantly
enhanced imaging characteristics:
- twofold sensitivity
- low noise.
Improvement
Next-generation back-illuminated CMOS
image sensor – Stacked Structure
- Compact Chip Size
- Larger logic circuit scale: 500k gates to
2400k gates
- Possible to optimize manufacturing
process separately for pixel and circuit
section
Ref: http://www.sony.net/SonyInfo/News/Press/200806/08-069E/
Ref: http://www.sony.net/SonyInfo/News/Press/201201/12-009E/
29. Technology: Camera
GPS
Lidar
Radar
Camera Ultrasound
There is a possibility of replacing short & Medium-range rader / laser sensors with
stereo / mono camera system.
Future Aspect
Ref: http://eecatalog.com/fpga/2013/02/07/driver-assistance-systems-with-the-power-of-fpgas/
Ref: http://www.aptina.com/solutions/automotive.jsp
30. Technology: Camera
GPS
Lidar
Radar
Camera Ultrasound
Driving becomes more difficult after sunset. Not only is visibility generally
reduced, but an even more serious challenge on unlighted roads are the
headlamps of approaching vehicles, which can dazzle a driver’s eyes. Currently,
there are in general two designs for night vision system: Passive and Active
Night Vision System
31. Technology: Camera
GPS
Lidar
Radar
Camera Ultrasound
Currently, Night Vision Systems (NVS) are only found out in a handful luxury cars:
e.g. Audi, BMW and Mercede Benz:
Feature
System Type
Price for NVS
Affected by Fog
Audi A8L
Thermal Image Sensor (Passive)
Pedestrian
Detection
Mercedes Benz CL55o
Illuminator + Image
Sensor (Active)
$2300
$2600
$3490
NO
NO
YES
YES
NO
Only warm objects, people and animal
All, including lane marks,
road signs, reflective road
markings, rocks, fallen
trees, etc
Daylight Visibility
Objects can be
viewed
BMW 750i
500 ft
300 ft
300 ft
Ref: http://electronics.howstuffworks.com/gadgets/automotive/in-dash-night-vision-system.htm
Ref: http://www.photonics.com/Article.aspx?AID=23154
Ref: http://www.onemotoring.com.sg/publish/content/onemotoring/en/motoring_buzz/motoring_accessories/bmw_night_vision.print.html
32. Technology: Camera
GPS
Lidar
Radar
Camera Ultrasound
For the last 25 years, number of IR image sensor pixel has been increasing at an
exponential rate, doubling every 19 months, which is in-line with Moore’s law growth
path:
Ref: http://www.sciencedirect.com/science/article/pii/S1350449510001040
33. Technology: Camera
GPS
Lidar
Radar
Camera Ultrasound
Price for uncooled thermal
camera is expected to drop
by as much as 60% in the
next 5 year (2013 to 2018)
Ref: http://image-sensors-world.blogspot.sg/2013/09/yole-reviews-uncooled-ir-imaging-market.html
Ref: http://www.sensorsmag.com/sensors-mag/sharply-lower-costs-uncooled-ir-sensors-may-spur-growth-7260
34. Technology: Ultrasound
GPS
Lidar
Ultrasonic sensors are used to assist
the driver in autonomous vehicle.
Radar
Video
Ultrasound
Radar and ultrasonic sensor
(Courtesy Analog Devices)
Variable gain amplifier used
in ultrasonic sensor
(USD12 each)
(Courtesy Analog Devices)
35. Technology: Ultrasound
GPS
Lidar
Radar
Video
Ultrasound
Cost Down: Advances in Integrated Circuit packaging potentially can drive down ultrasonics
component prices.
Embedded Ultrasonic sensors using
“Chip in Polymer” technology
Estimated to reduce cost by 30%
(courtesy of Fraunhofer Institute IZM)
Embedded ultrasonic sensors for
Automotive.
36. Technology: Ultrasound
GPS
Lidar
Radar
Video
Ultrasound
Parktronic option price on Mercedes E-class Assumptions:
Possible cost down of 5%Possible cost down of 5%
this year
due to advances in circuitfrom 2015
packaging
onwards due process
improvements.
1000
900
800
700
600
500
400
300
200
Price
100
0
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
(Courtesy of Motortrend)
2017
2018
2019
2020
37. Driverless Fleets
The Future: Vehicle-to-Vehicle/ Infrastructure Interactions
Vehicle to Vehicle (V2V) communication
(Courtesy of autoevolution)
Vehicle to Infrastructure (V2I) communication
V2V and V2I-I2V Communications
(Courtesy of Embedded System Technology)
Communication protocol is based on IEEE 802.11p
Ref: http://www.autoevolution.com/news/gm-begins-testing-of-vehicle-to-vehicle-communication-in-ann-arbor-48560.html
Ref: http://en.wikipedia.org/wiki/IEEE_802.11p
Ref:http://www.uwicore.umh.es/files/paper/2012_international/uwicore_IEEE_Communications_802.11p%20vehicle%20to%20infras
tructure%20communications%20in%20urban%20environments.pdf
38. Driverless Fleets
The Future: Vehicle-to-Vehicle/ Infrastructure Interactions
Data transmission improvement from Year 1999 to Year 2012
(802.11 Network standard)
Logarithmic increase in
data transmission speed.
10000
per data stream (Mbit/s)
Mbit/s (log scale)
1000
100
10
1
802.11a (Yr1999)
802.11b(Yr1999)
802.11g(Yr2003)
802.11n(Yr2009)
802.11ac(Yr2012)
(Courtesy of wiki/IEEE_802.11)
Ref: http://www.ruhr-uni-bochum.de/integriertesysteme/emuco/files/hipeac_trends_future.pdf
802.11ad(Yr2014)
40. Systems Integration (III)
Sensor data --- Central Processing Unit
What autonomous car
see making a left turn
(Courtesy of Robotics
Business Review)
Google car monitoring traffic
(Courtesy of CNN)
750MB of sensor data is
collected per second
(Courtesy of Robotics
Business Review)
CPU in Google Driverless car
(Courtesy of Robotics Business Review)
Ref: http://www.roboticsbusinessreview.com/article/google_car_data_hog_at_speeds_topping_2700gb_per_hour
41. Systems Integration (IV)
Improvements microprocessor speed with time
Improvement made in
1. Transistor count
2. Clock speed (MHz)
3. Power consumption
4. Perf/Clock (ILP)
(Instruction-level parallelism)
Our estimate for the number
of microprocessor used in
Google Car is at least three 2
GHz processor in parallel to
handle the 750 MB of sensor
data per second.
Ref: http://www.gotw.ca/publications/concurrency-ddj.htm
Ref: http://www.extremetech.com/computing/116561-the-death-of-cpu-scaling-from-one-core-to-many-and-why-were-still-stuck
42. Cost Down Estimates
Gasoline car
USD $20,000
Google car
USD $150,000
- USD $20,000
= USD $130,000
$140,000
Component Cost Down ~ 30% Annually
$120,000
$100,000
Year 2020
USD
$80,000
USD $10,706 + 20,000
= USD $30,706
$60,000
$40,000
$20,000
$0
Cost
2013
2014
2015
2016
2017
2018
2019
2020
$130,000
$91,000
$63,700
$44,590
$31,213
$21,849
$15,294
$10,706
43. Ongoing Projects / Models
Entrepreneurial Opportunities
Group/Player
Autonomous
vehicle/system/project
Stanford & Udacity, United States
Google Driverless Car
GM-Carnegie Mellon Autonomous Driving Collaborative Research Lab
Boss car
University of Berlin, Germany
Madein Germany vehicle
Technical University of Braunschweig, Germany
Stadpilot
Universität der Bundeswehr, Munich, Germany
MuCAR-3
VisLab, University of Parma, Italy
BRAiVE prototype
Oxford Mobile Robotics Group, UK
Wildcat
INRIA IMARA, Paris, France
Cybercar, cybercars2, LaRA project
Induct Mobility Solutions, France
Navia shuttle, CityMobil2 project
Griffth University, Intelligent Control Systems Lab, Australia
Cooperative driverless vehicles
Electronics and Telecommunications Research Institute (ETRI), Korea
ETRI Smart Transport RObot
Autonomous Solutions, Inc (ASI), USA
Vehicle Automation Kit
Singapore-MIT-Alliance for Research and Technology, Singapore
Future Urban Mobility
44. Entrepreneurial Opportunities
Autonomous Vehicle
Manufacturers
Increased demand for AV:
• Size, shape, function
customizable
• Open to all users: e.g. the blind,
the handicapped, elderly, children
• AV servicing industry.
Ref: KPMG
Components
Manufacturers
Increased demand for components:
• GPS, Radar, Camera, Lidar,
ultrasonic sensor etc.
• Computer hardware: e.g. high
performance ICs, storage, LCD,
wireless comm, Audio etc.
45. Entrepreneurial Opportunities
Software Developers
Info-Management Companies
• Artificial Intelligence model and
software act as brain of AV
• “Big Data” generated in V2V, V2I
has enormous analysis values
• Real time traffic monitoring data
can be sold to AV comm-network.
Ref: KPMG
Government
Social Services
• AV can be used in Car Rental, Car
Sharing, Auto-Taxi schemes.
Reduce operation cost and
ownership cost by 80%
• New insurance packages for AV.
• Free up lands for various purposes
46. Entrepreneurial Opportunies
Autononmous Car – Interior Design Companies
Change TVRest andOpening Concept
of Doors Relax
Reverse Entertainment Surfing
Windows SeatingtheEnabled
Workingwith Internet
Voice ControlSpace
on
Move
PSP Entertainment
Spacious Arrangement
47. Incentatives for Autonomous Car
Initial phase
Dedicated lanes for Autonomous cars
Tax break for autonomous car buyers
City state like Singapore implement
all autonomous cars only.
(Productivity gain and no more traffic jam)
48. Road Map for Autonomous Car
2012
• Google Driverless car project tests a fleet of self-driving
vehicles, including six Toyota Priuses, an Auti TT, and
three Lexus RX450h’s
• Mercedes, Cardillac, BMW, Auti, and Nissan models
include autonomous steering, braking, acceleration,
parking, lane guidance, accident avoidance and driver
fatigue detection
2014 - 2015
• Google’s fleet of fully autonomous Toyota Prius
hybrids will be released
2018
2020
2020
onwards
• Volvo, GM, Audi, Nissan and BMW have fully
autonomous cars
• Disrupt the conventional car industry!!!
Ref: http://www.carloan4u.co.uk/infographics/the-ultimate-car-of-the-future/