+ All Categories
Transcript

1

TI 資深汽車處理器應用工程師Rio Chan

進階駕駛輔助系統解決方案設計簡介

2

Designing Deep Learning Solutions for ADAS

Outline

• What is ADAS?

• What are the important peripherals for ADAS?

• What Deep Learning can do for ADAS?

• TI TDA4 ADAS design concept?

• TI TDA4 road maps?

• TI Deep Learning.

• TIDL Demos?

• What TI can support?

3

ADAS?

4

ADAS level

5

WHAT ARE THE IMPORTANT PERIPHERALS FOR ADAS?

6

What are the important peripherals for ADAS?

TI TDA4

https://www.ti.com/product/TDA4VM

7

What are the important peripherals for ADAS?

TI Radars

https://www.ti.com/sensors/mmwave-

radar/overview.html

8

What are the important peripherals for ADAS?

TI CAN solution - TCANx

https://www.ti.com/solution/drive-assist-ecu

9

What are the important peripherals for ADAS?

TI PGAx U-Sensor AFE

https://www.ti.com/solution/ultrasonic-

park-assist-sensor

TI LMGx DCDC boost https://www.ti.com/solution/automotive-

mechanically-scanning-lidar10

TDA4 ADAS DESIGN CONCEPT

11

TI Processor SDK | Single development kit for all families

12

Cockpit Cluster Amplifier Powertrain Camera Radar LIDAR

Sensor

fusion

100% code

compatibility

TI-RTOS

Gateway

Safety Security

System

Performance

DRA8x TDA4xAM6xAM2x

Jacinto’s DRA benefit for “compute” system

13

Jacinto

DRA

Jacinto’s TDA4 benefit for “ADAS” system

14

Jacinto

TDA

TI Jacinto™ 7 | Heterogeneous compute platform

15

Optimize entire platform around

programmer productivity on the

MPUs

Choose the right core

for the right job

Offload the majority of “work” to

specialized processors. Provide

tools & SW to manage complexity

Heterogeneous

Compute Capabilities

Image Signal

ProcessorDSP for

analytics

Deep Learning

Acceleration

Vision

Accelerators Display Engine

4K

Multi-Media

Accelerators

GPU

Dual-core

RF5 clusterDual-core

RF5 cluster

Jacinto™ automotive

processors

64b ARMv8

MPUs

Common high-performance chassis• High-bandwidth coherent SoC interconnect for main processor & memory

• Packet-based interconnect for latency-sensitive interfaces (Ethernet, CSI)

• Dedicated boot engine (early boot)

Memory

System

L3$

Safety• ASIL-B capable SoC

• Safety HW, SW & collateral

• ECC on all memories & buses

Security• Run-time and secure boot

• SoC-level firewalls, BIST

• Dedicated crypto HWAs

Automotive connectivity• CAN-FD, MOST-150, FlexRay, Ethernet AVB, USB 3.1

• GbE switch, PCIe Gen4 with switch

ASIL-D

Safety MCU

RTOS

Dual-core

A72 cluster

Dual-core

A72 cluster

Dual-core

A72 cluster

Dual-core

A72 cluster

More

Pro

cessin

g

Pe

rform

ance

Unique TI

innovation

Embedded Machine Learning Architectures

16

17

TDA4 ADAS Design concept

18

TDA4

Radar

ADAS Camera mirror

ADAS Domain controller

DMS / In-cabin Sensing

Surround view / Parking

Front Camera

19

20

2121

2222

2323

24

TDA4 ROADMAP

25

Jacinto Vision Analytics Portfolio

26

TI DEEP LEARNING

27

What Deep Learning can do for ADAS?

28

Edge inference software stack on TI SoC

29

TFLite RunTime ONNX RunTime TVM /Neo-AI-DLR

Cortex-AC7x + MMA

(Deep Learning Accelerator)

TIDL Optimizer: Layer fusion | Post Training Quantization | Range Calibration | Memory optimizations

TIDL RunTime: API | TIDL library on C7x & MMA

Model artifacts

30

TIDL RunTime

User ApplicationPython / C / C++

API | interpreter | scheduler

TFLite/ONNX-RT/Neo-AI-DLR

Jacinto 7 processor

Deep learning accelerator

-+ * =

C7x DSP with MMA*

ARM Cortex A72

ARM Cortex A72

IPC

Linux OS

CPU

TIDL Library

Open-source kernels

31

TI Deep Learning | SW Eco System

32

Training Inference

TIDL Importer

Deep Learning Application Execution

OpenVX

Framework

TIDL Inference

Training frameworks

TIDL Graph Compiler

TIDL Importer➢Enables development on Open frameworks

➢Model Parameters Quantization & Calibration

➢Layers Merging/Fusion

TIDL Graph Compiler➢Memory management, Network Execution

Plan/order

➢Models the device & Estimates performance

TIDL Host Emulation➢PC/Host emulation of Inference for offline

Validation and Accuracy benchmarking

TIDL Inference➢Abstracts embedded development

➢High Efficiency Inference Implementation

OpenVX Node ➢Deep learning Use case development

across cores (IPC, Pipelining )

Model Formats

PrototxtCaffemodel

Frozen-pb/TFLite

TIDL Host Emulation

What Deep Learning can do for ADAS?

33

Real-time 8x 2MP camera input demo on single ISP

34

• TDA4VM Camera performance as below.

• 8x 2MP cameras @ 30 fps

• Feeding to single TDA4VM device via CSI-2

• All camera images pre-processed by single

ISP in real-time• ISP still has headroom to process 2 more

cameras

• Flexible video processing sub-system

YouTube link: https://youtu.be/zsYbPPhO0t8

DMPAC: Depth and Motion Perception Accelerator

35

Stereo disparity map

Stereo Depth Engine: for each location <x,y> , estimates depth by

measuring disparity from 2 different views

Assign confidence score for each disparity output

Stereo

Depth

Engine

90 Mpix/s, <0.5WUp to 2MPix

Dense optical flow

Dense

Optical

Flow

150 Mpix/s, <0.5WUpto 2MPix

Dense Optical Flow: for each location <x,y>, estimates 2D

motion vector field given two images

Generate confidence score for each flow vector output

Object Tracking Structure From Motion (3D)

Moving Object Segmentation

Object distance measurementCollision warning, Segmentation,Path planning,….

4x 1MP @ 20 fps

disparity map per core

TDA4 EVM AND DEMOS

36

TDA4V Mid EVM Details

37

Modular Development System

➢Provides complete access to Jacinto™ 7 SoC

➢HW platform for SDK Validation

➢New SOM for each Jacinto™ 7 SoC Variant

➢On-board XDS110 JTAG for debug via CCS

➢XDS560v2 for advanced debug and trace

http://www.ti.com/tool/TMDSEMU560V2STM-U

➢ Processor SOM

➢ Common

Baseboard

Baseline Hardware Kit

Optional Expansion cards for application needs

Gateway Ethernet Switch

Industrial Expansion

Infotainment

Expansion CardFusion Application

Board

additional

expansion

cards plannedhttp://www.ti.com/tool/TDA4VMXEVM

Available Now

Jacinto™ 7 TDA4VM analytics and vision demos

38

Front camera analytics with TDA4 Smarter Parking with TDA4

Deep Learning

Based Forward

Camera Analytics

➢ Semantic Segmentation and Free space ID

➢ Multi-class Object Detection

➢ Custom DL networks built on TI Deep Learning

Libraries (TIDL) and running on C7x+MMA

8MP Front Camera

Perception and

Localization

➢ NCAP ready multi-class object detection using

TIDL Libraries and C7x+MMA

➢ Fusion with IMU and GPS for Localization

➢ Concurrent applications on one TDA4VM SoC

Surround View

with Analytics

➢ High-quality 3D Animated Car Model, overlays

for projected path and distance warning.

➢ Transparent car view using history

➢ Object Detection, Lane ID, Parking spot ID

using DSP based algorithms.

Auto Valet Park

and Surround

View

➢ AVP with Semantic segmentation, object and

parking spot detection based on TIDL

➢ Simultaneous Real-time 4x 2MP camera

Surround view with 3D car model.

➢ All code freely available in TI’s Processor SDK

Real-time 8-Camera

Input to TDA4

➢ Eight 2MP cameras feeding single TDA4VM.

➢ All camera images pre-processed by single

ISP in real-time

➢ All code freely available in TI’s Processor SDK

TDA4: Multi-Camera vision processing

➢ Ruggedized Development platform for

applications on TDA4VM SoC.

➢ Multi-sensor interfaces: camera, radar, lidar…

➢ Software compatible with TI EVM based on

Processor SDK

DesignCore Rugged

Vision Platform for

TDA4VM

Jacinto™ 7 TDA4VM YouTube Videos

39

Front camera analytics with TDA4 Smarter Parking with TDA4

Deep Learning

Based Forward

Camera Analytics

➢ https://youtu.be/F4KNWuTSJds

8MP Front Camera

Perception and

Localization

➢ https://www.businesswire.com/news/home/202

00107006173/en/

Surround View

with Analytics

➢ https://youtu.be/F4KNWuTSJds

Auto Valet Park

and Surround

View

➢ https://youtu.be/PER77CgoHWk

Real-time 8-Camera

Input to TDA4

➢ https://youtu.be/zsYbPPhO0t8

TDA4: Multi-Camera demo

➢ https://youtu.be/bOSmeej9_cQ

DesignCore Rugged

Vision Platform for

TDA4VM

WHAT TI CAN SUPPORT?

40

Software and tools - overview

41

TDA4 AutoSAR

42

TDA4 PinMux tool

43

https://e2e.ti.com/support/processors/f/processors-forum/927526/faq-ccs-tda4vm-pinmux-guide-for-jacinto-processors

TDA4 Cloud AI Tool: https://dev.ti.com/edgeai/

44

TDA4 Cloud AI Tool: https://dev.ti.com/edgeai/

45

TDA4 Cloud AI Tool: https://dev.ti.com/edgeai/

46

TI ADAS system of experienced partners

47

Camera Sensors, Modules

and Tuning Services

MCAL

AutoSAR Classic

Adaptive AutoSAR

SAFETY SOFTWARE, CERTIFICATIONOPERATING SYSTEMS AND SECURITY

ENGINEERING SERVICES

Algorithm Porting and Optimization

Hardware Design

System Testing

Driver Development

Surround and Auto Parking

DMS and In-Cabin Monitoring

Front Camera and Autonomous Drive

ADAS ALGORITHMS

Thank you

48

IMPORTANT NOTICE AND DISCLAIMERTI PROVIDES TECHNICAL AND RELIABILITY DATA (INCLUDING DATASHEETS), DESIGN RESOURCES (INCLUDING REFERENCEDESIGNS), APPLICATION OR OTHER DESIGN ADVICE, WEB TOOLS, SAFETY INFORMATION, AND OTHER RESOURCES “AS IS”AND WITH ALL FAULTS, AND DISCLAIMS ALL WARRANTIES, EXPRESS AND IMPLIED, INCLUDING WITHOUT LIMITATION ANYIMPLIED WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE OR NON-INFRINGEMENT OF THIRDPARTY INTELLECTUAL PROPERTY RIGHTS.These resources are intended for skilled developers designing with TI products. You are solely responsible for (1) selecting the appropriateTI products for your application, (2) designing, validating and testing your application, and (3) ensuring your application meets applicablestandards, and any other safety, security, or other requirements. These resources are subject to change without notice. TI grants youpermission to use these resources only for development of an application that uses the TI products described in the resource. Otherreproduction and display of these resources is prohibited. No license is granted to any other TI intellectual property right or to any third partyintellectual property right. TI disclaims responsibility for, and you will fully indemnify TI and its representatives against, any claims, damages,costs, losses, and liabilities arising out of your use of these resources.TI’s products are provided subject to TI’s Terms of Sale (https:www.ti.com/legal/termsofsale.html) or other applicable terms available eitheron ti.com or provided in conjunction with such TI products. TI’s provision of these resources does not expand or otherwise alter TI’sapplicable warranties or warranty disclaimers for TI products.IMPORTANT NOTICE

Mailing Address: Texas Instruments, Post Office Box 655303, Dallas, Texas 75265Copyright © 2021, Texas Instruments Incorporated


Top Related