New Architecture for Autonomous DrivingImpact of Autonomous Driving Roadmap ADAS architecture for automated driving is on the roadmap of many OEMs by 2020 About 45% the total ADAS

© 2018 IHS Markit. All Rights Reserved.

New Architecture for Autonomous Driving

June 4-5, 2018


2

C O N N E C T E D C A R

C O S T M A N A G E M E N T& P R O D U C T P L A N N I N G

A U T O M O N O U S C A R

A L T E R N A T I V E P R O P U L S I O N

In the Factory In the Car Supplier ManagementTechnology Roadmaps

ADASSensors

V2X CommunicationsAutonomous Driving

Hybrid & EVWireless ChargingCharging/Re-fueling Infrastructure

IoT & Cellular ConnectivityInfotainment

TelematicsMedia Integration

Smartphone & AppsAdvertisingNavigationWearables

Cyber Security

TECHNOLOGY Defining the Future

of Automotive


Vehicle production rises steadily but slowly

3

2.0% CAGR (2017 – 23)


Average value of electronic systems per car to top $1650 by 2023

4

6% CAGR (2016 – 23)


Electrification, automated driving and connectivity fuel for the automotive semiconductor growth

5

7.7% CAGR (2017 – 23)


AI, Machine Learning, Neural Net, Deep LearningMajor differences in the same “intelligent” family

AI ML NN DL

Code Data

Tra

ditio

nal

Alg

ori

thm

s


AI in Automotive

Security

HMIVoice/Gesture

Driver Monitoring

Diagnostic

Powertrain

From ADASTo Autonomous


Implication of AI and Deep LearningMajor advantages in comparison with traditional machine vision

• Assumptions:

> New silicon solutions will be developed with focus on AI algorithm

> The functional safety aspect will be addressed by the entire supply chain

• Deep learning can:

> Allow detection and recognition of multiple object ➔ improve perception

> Perform semantic analysis of the area surrounding the vehicle

> Reduce development time of ADAS and IVI systems (once DL is in steady-state)

> Reduce the power required compared to the same operation w/ traditional algo

• Deep Learning needs help

> Recognition/Prediction of actions and Fusion - Bayesian Net and other stochastic algorithms may complement DL in the run to autonomous cars (L4-L5)

• Required precondition:

> Telematics will be broadly deployed to: 1) enable gathering of “real” patterns and data for training 2) allow over the air system update and security

8


Extra Requirements for Deep Learning in ADAS & AV

• DL in ADAS for Autonomous functions requires in-vehicle HW:

> Latency: for active function system needs to react in less than 70-80ms

> Deep Learning offer deterministic latency also for “noisy” input from sensors

> Performance: TFlop/TOP/TMAC is barely the minimum

• Power:

> Individual sensor subsystems need to stay in the power budget of 4W;

> Sensor Fusion ECUs might allow targets up to 15-20W or more. Some OEMs expect already they need to find a trade off if no silicon is available and performance needed.

• Backhaul and data storage infrastructure:

> Connectivity (IoT) is a need to:

– Store training data and vehicle parameters.

– Update/Upgrade the system

• Data acquisition is a challenge for validation and test: mix Real & Synthetic data

• Safety is the biggest uncertainty to have autonomous car based on AI.

9

Standardisation is a must have


Dynamic of ECU and Features balanceCost will define the strategy and implementation timeframe

10

• Number of ECUs (Average):

• in Low and Mid vehicle segment continue increasing.

• Premium shows already consolidation because close to the “limit”.

• Feature:

• growth in number expected in all segments, driven by ADAS and IVI

• Costs

• OEMs tend to maintain the ECU cost at a constant level adding value and taking advantage of integration.


Old generation: E-segment with 14 ADAS ECUs

RADAR

RADAR

RADAR

FIRNV CU

RADAR

RADAR

CAMCAM ECU

CAM

CAM

CAMSV PARK

ECU

PARK ECU

CAN

11

BMW 7-Series 2009

Average Semiconductor BOM: $618


Advancement of features on new platform

RADAR

RADAR

RADAR

FIRNV CU

RADAR

RADAR

CAM

CAM

CAM

CAMSV PARK

ECU

PARK ECU

CAM

Ethernet

FUSION ECU

12

BMW 7-Series 2015

Average Semiconductor BOM: $563 (-10%)Average Electronics Value: $1742


System Architecture on Audi A8

RADAR

RADAR

RADAR

FIRNV CU

RADAR

RADAR

CAM

CAM

CAM SmartCAM

ADAS DC

Audi A8 2018

CAM

LIDAR

DM Cam

+12 ultrasonic sensors

IHS Markit Technology - ADAS architecture strategies

13


Possible System Architecture in 202X

CAMMPU

RADAR

MPU

CAM

CAM

CAM

RADAR

RADAR

FIR

MPU

RADAR

LIDAR

LIDAR

LIDAR

LIDAR

V2X

V2X

CAM

RADAR

Autonomous car

14

DC1HW1+SW1

DC2HW2+SW2

Actuation ASIL-D


Typical ADAS architecture requirements

ADAS Module Avg. per L3 Avg. per L4 Avg. per L5

Sensor Fusion 1 2 2

Exterior Camera 5 8 8

Interior Camera 1 1 1

Short/Mid-range Radar

4 6 6

Long-range Radar 1 2 2

Long-rangeLIDAR

1 1 1-2*

Short-range LIDAR

2* 2-4* 4

15

*Architectures based on existing pilot car platforms from BMW, Volvo, Audi, Nissan..

242213

$ 2200 $ 6300 $ 9400


Impact of Autonomous Driving Roadmap

ADAS architecture for automated driving is on the roadmap of many OEMs by 2020

About 45% the total ADAS cost on premium models is accounted for the softwareParticularly on A8 and Model S the average software value per ADAS module (excluding ultrasonic sensors) is $90 and $77, above today´s average.

Functional safety standards: The control units (HW & SW) supporting AD functions should also comply ASIL-B to ASIL-D. Redundancy is a must.

High-performance software blocks: Front smart cameras, Domain Controllers & LIDARs

LIDARs and Domain controllers for instance come with a cost that ranges from 500 for a LIDAR to 850-1000$ for the domain controller.

16

* AD=Autonomous Driving


ADAS classification by passive warning and active control

*The results of Tesla Model functions will be updated in the final deliverable

17


Software value driven by performance and safety

18


Battle in the SoC space: heterogeneous architecture

19

• GPU, TPU, CPU, IPU….❑ Heterogeneous architecture!

❑ DLTOPS and DMIPS/MHz

• Cost of SoC over time❑ Impact on scaling and tech-node

❑ Ensure <16nm and high-volumes

• SW vs HW❑ Platform scalability

• Go to market strategy❑ L2-L3 “now” or w/ L4-L5 “later”

0

5

10

15

20

25

30

35

Nvidia Xavier Mobileye EyeQ5 Renesas R-Car Other

[nm] DLTOPS [W]

Major SoC suppliers in automotive supporting AI

© 2017 IHS MarkitSource: IHS Markit


Performance and power: energy efficiency;…and safety

20

• In datacenter the energy efficiency is directly correlates to the cost➔ Power consumption is critical for auto too but compromise are acceptable

• Let´s consider the overall system power consumption ➔memory is about one order of magnitude more than SoC➔ embedded volatile memory is preferable for performance and power

• If >100 TOPS is the target in L4-L5 type of vehicle,….by when?➔ OEMs´ and suppliers´ strategy might differentiate

• Challenge for AI&DL: Deterministic behavior and ISO26262


Something probably needs to change….

21

TÜV SÜD and DFKI to develop “TÜV for Artificial Intelligence” The German Research Center for Artificial Intelligence (DFKI) and TÜV SÜD are launching a joint

project to certify systems based on artificial intelligence (AI) used in autonomous driving and

develop a ‘roadworthiness test’ for algorithms

Intel MobilEye - Overview of the Plan We believe that it is important for the automotive industry to collaboratively establish a

methodology and standard for safety validation in partnership with global standards–bodies and

regulators. The United States is among the countries leading the way with pending self-driving

vehicle legislation and new USDOT Automated Vehicle Guidelines, making this a perfect time to

engage in these collaborative next-step discussions.

Our proposed model provides a detailed, practical, and efficient solution for designing and

validating an AV system that results in drastically improved safety. Here is an outline of the next-

step areas we believe merit attention and the solutions we propose: …..


Thanks for your attention

[email protected]

Executive Director Transformative TechnologyAI and Automotive architecture

URL:https://technology.ihs.com/Research-by-Market

22

mailto:[email protected]

https://technology.ihs.com/Research-by-Market

Documents

New Architecture for Autonomous DrivingImpact of Autonomous Driving Roadmap ADAS architecture for automated driving is on the roadmap of many OEMs by 2020 About 45% the total ADAS