MATLAB을 활용한 자동차 레이더 개발

서기환

© 2015 The MathWorks,1 Inc. Agenda

▪ Different Usage for Modeling

▪ Radar Design Workflow – DSP Design and Simulation – RF/Antenna Modeling

▪ High Level Simulation with Probabilistic Model

2 Agenda

▪ Radar Modeling with Fidelity Control

▪ Radar Design Workflow – DSP Design and Simulation – RF/Antenna Modeling

▪ High Level Simulation with Probabilistic Model

3 Two Personas using Automotive Radar Sensor Models

Radar Designers Radar Users

Detections Detections Radar Sensor Radar Sensor World / Tracks World / Tracks Truth Truth IQ Data …

Lidar Sensor RF Signal Antenna Front-End Processing Camera Sensor

Ultrasound Sensor

4 Radar Modeling and Simulation

RF/Antenna Modeling DSPDSP AlgorithmAlgorithm ModelingModeling

Target Classification

Radar Signal Antenna/Array/RF Processing

Tracking and Sensor Fusion Scheduling and Radar Control

5 Radar Modeling and Simulation

Probabilistic Model

Target Radar Sensor Classification

Tracking and Sensor Fusion

6 Automotive Radar Sensor Models

DSP algorithm Model RF/Antenna Model Probabilistic Model Engineers Radar Designers Radar Designers Radar Users

Analog-mixed Signal Sensor Fusion, Model usage Radar Algorithms Simulation Controller

Outputs IQ Data or Detections IQ Data Detections or Tracks

Benefit - Highest Fidelity Simulation Speed

Disadvantage - Simulation speed Can’t Access IQ Data

7 Agenda

▪ Radar Modeling with Fidelity Control

▪ Radar Design Workflow – DSP Design and Simulation – RF/Antenna Modeling

▪ High Level Simulation with Probabilistic Model

8 Radar Simulation and Modeling Architecture

Waveform Transmit Transmitter Generator Array Pt λ Gt

Radar 2 Environment, 푃푡퐺푡퐺푟휆 휎 Scheduler and Targets, & 푃푟 = 2 2 3 Interference (L, 흈, 푹 푹 ) Tracker 4휋 푅푡 푅푟 퐿 풓 , 풕

Receive Signal Receiver Array Processing Gr Gr

▪ Functions for calculations and analysis ▪ Apps for common workflows ▪ Parameterized components for system modeling ▪ Easy path to increased fidelity for antenna and RF design ▪ Code generation for deployment

9 Radar Model to Simulate High Fidelity Raw Data

Mixed-Signal Algorithms Antenna, Antenna arrays Continuous & discrete time , beamsteering, type of element, # elements, configuration MIMO • Simulink • System Toolbox • Antenna Toolbox • DSP System Toolbox • Communications System Toolbox • Phased Array System Toolbox • Control System Toolbox • DSP System Toolbox

RX LNA ADC Signal Processing Tracking & • SensorSensor Fusion and Fusion Tracking Toolbox

PA DAC • Communications System Toolbox TX • Phased Array System Toolbox Channel • Phased Array System Toolbox interference, clutter, noise • RF Blockset • Signal Processing Toolbox • RF Toolbox • Wavelet Toolbox

RF Impairments Waveforms & Resource Scheduling frequency dependency, non-linearity, noise, mismatches 10 Path to Higher Fidelity

▪ Extend model fidelity over project evolution ▪ Simple interface to replace off-the-shelf components with custom ones

Antenna element Target model Propagation model RF signal chain

Ideal elements Point target Free space Baseband

EM solver with mutual Synthesized backscatter Line of sight atmospheric RF components coupling (angle & frequency) effects

Measured pattern import Measured return (angle & Multipath, terrain and ducting Analog-mixed simulation frequency) effects

11 DSP Algorithms for Radar Systems

Beamforming Detections

12 Design

or

Replicate to build array Design subarray with desired fidelity

Assess resulting pattern

13 Beam Steering

14 Synthesizing an Array from a Specified Pattern

▪ Introduce optimization workflow

https://blog.naver.com/matlablove/221205030640 https://kr.mathworks.com/company/newsletters/articles/synthesizing-an-array-from-a-specified-pattern-an-optimization-workflow.html

15 Hybrid Beamforming

▪ 4 subarrays of 8 patch antennas operating at 66GHz → 4x8 = 32 antennas ▪ Digital beamforming applied to the 4 subarrays (azimuth steering)

▪ RF beamforming (phase shifters) applied to the 8 antennas (elevation steering)

Beamformers (array and subarray)

4 subarrays

Array pattern Subarray weights 16 RF Front End Modelling using Circuit Envelope

▪ Direct conversion to IF (5GHz) and superhet up-conversion to mmWave (79GHz) ▪ Non-linear impairments such as IP2, IP3, P1dB. ▪ Power dividers (e.g. S-parameters) ▪ Variable phase-shifters

17 Radar Waveform Analyzer

18 Sensor Array Analyzer

19 Radar Equation Calculator, RF Budget Analyzer

20 Model FMCW at mmWave Frequencies

Waveform Transmitter Generator Transmit Array

Signal Processing Receive Receiver Array

21 Model FMCW Radar – RF Front-End

Transmitter

or

22 Model FMCW Radar – Transmit Array Antenna

Transmit Array

23 Model FMCW Radar – Signal Processing

Signal Processing

24 Visualizing Radar and Target Trajectory

25 Automated Driving Simulation with IQ-level Radar Signal

26 Simulating Micro-Doppler Signatures

Micro-Doppler for Pedestrian Bicycle

27 Pedestrian Micro-Doppler with and without Parked Vehicle

Micro-Doppler for pedestrian (only) Micro-Doppler for pedestrian and parked vehicle 28 Pedestrian and Bicyclist Classification Using Deep Learning

29 Pedestrian and Bicyclist Classification Using Deep Learning

30 Increasing Angular Resolution with MIMO Radars (Virtual Array)

2 Tx 4 Rx

Two options: Increase number of receive elements or perform signal processing

31 MIMO non-MIMO Non-MIMO MIMO

Simulation

Scene Live Data

32 Agenda

▪ Radar Modeling with Fidelity Control

▪ Radar Design Workflow – DSP Design and Simulation – RF/Antenna Modeling

▪ High Level Simulation with Probabilistic Model

33 Virtual Driving Scenarios with Radar Sensor

Scenes Cuboid 3D Simulation

Testing Controls, sensor fusion, planning Controls, sensor fusion, planning, perception

Authoring Driving Scenario Designer App Unreal Engine Editor Programmatic API (drivingScenario) Sensing Probabilistic radar (detection list) Probabilistic radar (detection list) Probabilistic vision (detection list) Monocular camera (image, labels, depth) Probabilistic lane (detection list) Fisheye camera (image) Lidar (point cloud)

34 Synthesize Radar Sensor Data

Simulate Radar Sensors in 3D Environment ▪ Extract the center locations ▪ Use center location for road creation using driving scenario ▪ Define multiple moving vehicles ▪ Export trajectories from app ▪ Configure multiple probabilistic radar models ▪ Calculate confirmed track

Automated Driving ToolboxTM

35 Design trackers

Multi-Object Tracker

Association & Tracking Detections Track Tracks Filter Management

From various sensors at various update rates

▪ Multi-object tracker ▪ Linear, extended, and ▪ Global Nearest Neighbor (GNN) tracker unscented Kalman filters ▪ Joint Probabilistic Data Association (JPDA) tracker ▪ Particle, Gaussian-sum, ▪ Track-Oriented Multi-Hypothesis Tracker (TOMHT) and Interacting Multiple ▪ Probability Hypothesis Density (PHD) tracker Model (IMM) filters

Automated Driving ToolboxTM Sensor Fusion and Tracking ToolboxTM

36 Design Multi-Object Trackers

Multi-Object Tracker Extended Object Tracking ▪ Design multi-object tracker ▪ Design extended object trackers ▪ Evaluate tracking metrics ▪ Evaluate error metrics ▪ Evaluate desktop execution time

Sensor Fusion and Tracking ToolboxTM Automated Driving ToolboxTM

37 Design track level fusion systems

Vehicle 1

Multi-Object Track Detections Tracker Fuser Tracks

Vehicle 2

Multi-Object Track Detections Tracker Fuser Tracks

38 Track-Level Fusion

Track-to-Track Fusion for Automotive Safety Applications

Occluded Parked vehicles vehicle fused observed by from vehicle 1 vehicle 1 Occluded Pedestrian pedestrian observed by fused from vehicle 1 Pedestrianvehicle 1 observed by vehicle 2

Sensor Fusion and Rumor control: the fused TM Tracking Toolbox track is dropped by vehicle 1 TM Automated Driving Toolbox because vehicle 2 is coasting and there is no update by vehicle 1 sensors 39 Radar System Modeling for Perception

Antenna/RF Waveforms Tracking/ Sensor Fusion Scenario Generation

Spatial Signal Processing Environment Deep Learning Code generation

HDL

40 Key Takeaways

▪ Choose the Right Modeling Method – You can control the fidelity

▪ Start Your First Radar Design with Various Apps

▪ High Level Simulation with Probabilistic Model – Tracking – Control

41 Thank You!

42