Mobileye s Advanced Driver Assistance Systems and Technology for Autonomous Cars For a Safer World

Transcription

1 David Oberman David Oberman was born in the UK and made Aliya to Israel when he was 18 years old. After BA and MA studies in East Asian Studies at the Hebrew University of Jerusalem, David received a Japanese Government scholarship for further studies at Hokkaido University in Japan. After serving as a lecturer in East Asian studies at the Hebrew University, David formed his own company called Israel Japan Enterprises, which served as a commercial bridge between Israeli and Japanese companies for 10 years. For the last 18 years, he has managed sales of Israeli high-tech solutions, concentrating on data communications products and SOC camera solutions for mobile telephones, and for the last 8.5 years has been Director of Sales of Mobileye's Advanced Driver Assistance Solution (ADAS) products in Asia, Australasia, Latin America and Africa. During this time, David and his sales team have built up a distribution network for Mobileye s Collision Avoidance Products of some 130 distributors in over 50 countries, in the process saving many thousands of lives and preventing many tens of thousands of injuries. David and his wife Yael, a PhD and clinical psychologist, live in Jerusalem.

2 Mobileye s Advanced Driver Assistance Systems and Technology for Autonomous Cars For a Safer World David Oberman, Director of Sales, Mobileye Vision Technologies Ltd.

3 AGENDA 1. Mobileye Snapshot Company and Technology 2. ADAS OEM 3. ADAS Aftermarket 4. ADAS and Fleet Management Systems 5. Shield+ Blind Spot Solution for protecting Vulnerable Road Users 6. Smart City/Big Data Solutions 7. ADAS Regulations Worldwide 8. Technology for Autonomous Vehicles 9. Alliances for Self-Driving Cars

4 is the global pioneer and leader in the development of collision avoidance and autonomous driving systems based on artificial vision technology 100+ distributors globally 650+ employees worldwide Operating in 48 countries Adopted by most of the world's major auto manufacturers The biggest IPO ever for an Israeli company in the US market ($890M, in 2014). Publicly traded on the NYSE: MBLY

5 Mobileye Lane Detection, Vehicle Detection and Pedestrian Detection

6 273 Models 215 Strong OEM Penetration and Contracted Growth 160 Growing Number of Models* *As of July 2015.

7 15 million+ vehicles with Mobileye Technology Leading ADAS Supplier to the OEMs

8 Strategy: Seeking to Work with All Automakers SoC & Software Design Packaging & Integration Implementation

9 Mobileye EyeQ Processor EyeQ4 Performance (Mhz) 512X 28nm FD-SOI EyeQ3 64X 40nm EyeQ2 6X EyeQ1 90nm 180nm 1X Production Year

10 EyeQ1 EyeQ1 EyeQ2 EyeQ2 EyeQ3 180nm 180nm 90nm 90nm nm 2015 First Camera/Radar Fusion First Bundling of LDW, IHC, TSR First Pedestrian AEB First Camera Only ACC and Traffic Jam Assistant First Camera Only Full Auto Braking (AEB) First Animal Detection First Camera Only FCW First Camera Only AEB (partial braking)

11 Aftermarket ADAS Solution The same Advanced Collision Avoidance technology embedded in OEM models can also be fitted in any vehicle as an aftermarket solution. Lane Departure Warning Forward Collision Warning/Urban FCW Headway (Distance) Monitoring and Warning Pedestrian and Cyclists Collision Warning Automatic Headlamp Control Speed Sign Recognition

12 ADAS Solutions Single Units Shield Plus Passenger Vehicle Long Vehicle (Bus/Trucks/Trams)

13 Philip Morris Ukraine - Crash rate decreased by 58% Philip Morris equipped in Ukraine 350 passengers cars from its commercial fleet. Former crash rate was around 200 accidents/year in 2016 it was down to only

14

15

16

17 Fleet Management Sector Partners

18 Average incidents Headway alert 40% reduction Series1 HMW Series2 LDW Series3 FCW Forward collision 50% reduction Lane departure 80% reduction Weeks

19

20 The Problem *Increasing number of Vehicle-to-Vulnerable Road Users injuries and fatalities *Infrastructure deficiencies

21 Our Solution Many common accidents occur when a large vehicle is turning onto a side road and a pedestrian, cyclist or motorcycle cannot be seen. To address this issue, Mobileye Shield + was developed

22 Mobileye Shield+

23

24 Actual NYC Police Reports, show numerous cyclist injuries along North and South side of Fort Greene Park

25

26 Street View Unprotected bike lane Dangerous potholes force cyclists into traffic Unprotected bike lane Driver s open door into bike lane No protection from traffic

27 We can then make intersections safer Curb Extensions Medians & Cyclist Cut-Throughs Landscape Islands

28 Despite our success in reducing passenger vehicle occupant deaths by 33 percent in the past decade, in the past few years we have seen rising numbers of pedestrian and bicyclist injuries and fatalities related to collisions with motor vehicles. We must better protect people on foot and bicycle... so that even more Americans will feel comfortable enough to take advantage of walking and biking.

29

30 Boris Johnson, Mayor of London London Welcomes Mobileye Road Safety Technology Ex-Mayor Boris Johnson: It s fantastic to see that cutting edge technology developed in Israel is being utilized to help make London s roads safer.

31 Regulation

32 Government Mandate for Advanced Driver Assistance Systems

33 Automatic Emergency Braking Laws for Buses and Trucks: New Government Incentive for ADAS Sticker for Buses

34 The Korean regulation is for AEBS in new trucks and buses

35 Government Incentive for Advanced Driver Assistance Systems 70% subsidy for companies that implement MOBILEYE In their fleet In addition 3 insurance companies offer 10% discount to participating fleets

36 Taiwan The Taiwanese government promotes use of ADAS (LDW/FCW) in all inter-city buses. The government provides 49% subsidy.

37 STATE OF ISRAEL MINISTRY OF TRANSPORT AND ROAD SAFETY Israel's Ministry of Transport issued a mandate requiring ALL VEHICLES WEIGHING 3.5 TONS OR MORE TO INSTALL COLLISION AVOIDANCE SYSTEMS WITH FCW AND LDW This mandate includes vehicle models from 2012 and went into effect November 1 st 2016.

38 Government Incentive for Advanced Driver Assistance Systems As of August 2013, the Israeli Ministry of Transport and the Tax Authority grant tax incentives for vehicles equipped with safety systems. Vehicle importers receive an import tax reduction from the Israel Tax Authority for the cost of the Mobileye system, and are therefore able to supply new cars to the Israeli market with an Aftermarket Mobileye system installed, with no cost to the importers.

39 Israeli Government Incentive Requirements to Determine Safety Equipment Level POINTS SYSTEM Lane Departure Warning System Headway Monitoring Warning (Distance Alert) Adaptive Cruise Control Pedestrian Collision Warning Automatic Emergency Braking (AEB) Blind Spot Information System 7 Air Bags + (industrial 6) Reverse Cameras Intelligent High-Beam Control Seat Belt Buckling Sensors (front & back)

40 Israeli Government Incentive Safety Systems and Points to Determine Safety Equipment Level Points Required Safety Equipment Level Required Amount of Air Bags Private Industrial , , ,800 Tax Discount (NIS) , , ,250

41 China The Chinese Ministry of Transport carried out a trial of Mobileye in 300 trucks and buses in The results were excellent no crashes and a big reduction in the number of warnings The Ministry of Transport and Mobileye held a seminar and press conference on May 16-17, 2016, to report the results of the pilot. GB7258 new traffic safety regulation in China, will mandate LDW and FCW in all buses over 11 metres long, and later trucks. This law is now at Request for Comment stage and is expected to be published on

42 United States The Chairman of the National Transportation Safety Board (NTSB), Mr. Christopher Hart, announced that use of ADAS can prevent or mitigate around 80% of forward collisions The 10 leading car manufacturers have guaranteed to supply all new cars with ADAS by 2022.

43 Autonomous Vehicle

44 Mobileye co-founder Professor Amnon Shashua drives an Audi A7 with Mobileye functions

45 Functions of the Audi A7 test vehicle

46 Free Space and Road Scene Interpretation

47 Mobileye Technologies - Computer vision surround-view environment model Breaks down to the following tech sub-domains : Movable objects detection Driving Path sensing Drivable area boundaries Scene semantics extraction Covering static scene elements, the driving paths and objects within it ; i.e. traffic regulation indicators and elements, target intention (Ped orientation, Vehicles turn/brake indicators), Road boundary types, etc. - Sensor Fusion Drivable area and objects, low and high level fusion. Lidar and Radar. - Driving Policy and planning - Mapping and localization

48 Surround sensing - base proposition Front Fisheye 150 o Narrow 28 o Main 52 o 8 Cameras Features Tri-Focal Narrow 28 Mid 52 Fisheye 150 Front UHR 120 deg 7.4 Mp VD, PD, lane detection, HPP, Freespace, GO, TSR, TFL, HLB, Hazards, CCI, Crossing Objects, Arrows on Road Pillar 75 o Pillar 75 o Wing 100 o Wing 100 o Rear 52 o corner 75 Wing 100 Rear 52 VD at any angle Pedestrian Detection Freespace Lanes detection VD at any angle Pedestrian Detection Freespace VD at any angle Pedestrian Detection Lanes Detection TSR

49 Surround Vision

50 3DVD 50

51 3DVD 51

52 Parking Application Vehicle vision sensor identifies empty parking spots, resulting in dynamic map of free parking spaces 5 3

53 Free space drivable area boundaries 54

54 General objects detection 55

55 Hazards and debris detection

56 Road vertical geometry 5 7

57 Path Planning 58

58 Semantic Points Vocabulary 59

59 Complex scenes road semantic description HL host left HR host right NLL left mark of next lane left NRR right mark of next lane right NRL Left mark of next lane right NLR right mark of next left lane NLL Lane-Disjoined-right NRR HL / NLR HR / NRL 6 0

60 61 Scene semantics : TSR

61 Scene semantics : TFL, on-road traffic signs TFL detection Complicated structures State of the art super resolution TFL at night Stop line detection Arrows on the road Naive ego lane assignment RLW & SSW application 62

62 Semantic lanes information 6 3

63 Junction semantics : Braking point calculation 64

64 65 Scene semantics : Road-bump Detection appearance based

65 Scene semantics : Road Surface type 66

66 Road boundaries semantics - day Objects Flat Step (Curb( vertical Guard-rail Concrete barrier # o 67

67 Road boundaries semantics - night Objects Flat Step (Curb( Vertical Guard-rail Concrete barrier # o 68

68 Object semantics : intention indications, awareness assessment. Vehicle light indicators Pedestrian awareness On curb/road? Facing host?

69 THREE PILLARS OF AUTONOMOUS DRIVING SENSING Environmental Model 360 awareness Autonomous Driving REM (Roadbook) Road Experience Management Localization at High Accuracy (10cm) Drivable Paths DRIVING POLICY (Planning) Negotiating in a multi agent game

70 Autonomous Vehicles Mobileye believes that there are 3 parts to Autonomous Driving 1. Sensing 2. Mapping - REM 3. Policy

71 REM overview Mapping and localization engine for high autonomy support. End to end solution : harvesting client / aggregating server / self localizing client Economics : Requires only basic, low cost physical sensors Requires only single camera, and standard GPS, leaning on ME s computer vision and machine learning core competencies. Real-time geometrical and semantic analysis minimizes the transaction communication bandwidth to the aggregating server. (10KB/Km) Making mass market low-end vehicles natural and relevant harvesting clients resulting in ultra high refresh-rate of the map. By design does not depend on acquiring accurate global-coordinates localization to support AD. Yet may be used to produce accurate global positioning (<0.3m) of landmarks in the environment.

72 REM System Ingredients: On-board Camera (Client) Sparse 3D Dense 1D Crowd sourced Visual Landmarks

73 Landmark Density - Germany

74 Driving Policy Example: Negotiating entry to a Roundabout Blue = aggressive drivers, Green = courteous drivers Confidential

75 Example: Negotiating entry to a Roundabout After few iterations After many iterations Confidential

76 Mobileye Alliances for Autonomous Vehicles Audi VW Nissan-Renault General Motors BMW+Intel Delphi HERE Lucid Motors Others (To be Announced)

77 Thank You. Please Drive Safely.