Frequency-Modulated Automotive Lidar

1 Copyright 2018 XilinxFrequency-Modulated Automotive LidarJim CurryVP of ProductBlackmore Sensors and Analytics, Copyright 2018 XilinxFMCW Measurement TheoryPulse BWDetector BWDigitizer BWPeak PowerLimiters to rep rate, resolution, and rangeOptical BandwidthElectronic BandwidthPeak PowerFMCW Uncouples Traditional System LimitationsMost 3D Lidar systems are pulsed, direct detect and adaptive technique which provides compelling advantagestrepPowerTimeDirect DetectiontrepPowerTimeHeterodyne Detection = ChirpsConstant Power / Frequency ModulatedSuperior Resolution, Dynamic Range, Single-Photon SensitivityDoppler MeasurementMulti-Return ResolutionEye SafetyLow Cost Copyright 2018 XilinxFMCW Measurement TheoryLORXO ptical FrequencyTime = The properties of light are leveraged to make a physically superior measurementFFT to get range profile Power (dB)Range = 2 The optical bandwidth, B, can be increased electronically to meet nearly any resolution requirement Result is a beat note with frequency that is digitized and closelyspaced targetsThe LO/RX mixing process is beat note is measured, not the RX signal floor of the LO(very stable!)

2 Copyright 2018 XilinxNo Compensationw/ Doppler CompensationDoppler ChallengesStatic Imaging of Trees, Slight BreezeIn the infrared band, Doppler shifts are ~1 MHz per m/s. This is very challenging to deal with compared to radar (~1kHz per m/s). Copyright 2018 XilinxSimultaneous Range and VelocityStationaryApproaching Receding Pulsed Lidar systems do not simultaneously measure range and velocity Radar systems measure velocity but have poor spatial resolution Coherent (FM) Lidar combines high-resolution spatial and velocity measurements For perception, instantaneous velocity is very valuable Faster and more accurate scene segmentation Velocity content of tracked objects helps classify them (micro-doppler signatures)5+ Direction Copyright 2018 XilinxSimultaneous Range and Velocity6 StationaryApproaching Receding+ Direction6150m Copyright 2018 XilinxInterference Rejection AM Lidar pulses look exactly the same Lidar -to- Lidar interference cited as a pain point Sunlight elevates noise floor of AM measurements, reducing Lidar sensitivity in the daytime Blackmore s FM Lidar operates interference-free No issues with sunlight No issues with other cars sensors No issues with self-interference Copyright 2018 XilinxLong-Range Performance Sensor LocationIn-N-Out Burger SignStructure @ 400mGives autonomous systems plenty of range for smooth braking Copyright 2018 XilinxA Scalable SolutionTechnology leverages semiconductor foundries+Low-power continuous-wave laser compatible with integrated photonics, including optical phased arraysNon-Mechanical Beam Scanning Copyright 2018 XilinxEExample FM Automotive Lidar data Copyright 2018 XilinxSingle-pass mappingHeight (Altitude)

3 ColorationOne pass in one directionOne pass in each directionMoving points stripped out using velocity field Copyright 2018 XilinxHigh-Speed PerformancePerformance Maintained at Highway SpeedsHeight (Altitude) ColorationMoving Points Removed Copyright 2018 XilinxPoint-by-Point VelocityDirect velocity measurement enables segmentation and tracking of moving objectsFoot trafficVelocity Coloration Copyright 2018 XilinxMotion-compensation of velocity dataMicro-doppler data reveals wheel-spinPrecise, high-dynamic range velocity data Copyright 2018 XilinxQuestions?Height (Altitude) ColorationReflectivity Shading No saturation or blooming artifacts from bright reflectors Extremely high damage threshold Easily measures reflectivity differences from pavement markingsHigh-Dynamic-Range Reflectivity Data Copyright 2018 XilinxHigh Spatial PrecisionEnables detection of even minor pavement defects Copyright 2018 XilinxSee through, detect, and identify obscurantsVelocity coloring. Grey points=water surface, colored points=mistThin layer of sufacemist over pool Copyright 2018 XilinxWing-Walk in Doppler FieldQuestions?

4 Jim CurryVP of ProductBlackmore Sensors and Analytics.