Measurement Solution for new Radar Microcontroller V1.01 2015-12-03
New Vehicle Architecture Technology Change E-Drive Cloud computing Autonomous driving Connectivity Security ECU less age Some ECU age ECU Network age Near Future 2
New Vehicle Architecture ECU Technology Prognose 3 (Source: INTEL Automotive Division)
New Vehicle Architecture Impact on: DCU Micro Controller / OS / Bus Interfaces Micro Controller + ECU MC-Interfaces: PCIe Aurora with up to 25 GBit/s ECU OS + Adaptive XCP standard update necessary: No fixed mapping: Signal<> RAM address Bus Interfaces Eth. => Tap Mode + 4
ECU RAM Access XCP- ECU Access Datarate XCPonCAN 0,5 MBit/s XCPonFR. JTAG XCPonEth. Factor 50.000 DAP2 PCIe Aurora 25 GBit/s 5
High speed ECU (RAM) measurement XCP Dataflow via Serial Bus Interfaces or Debug-Interface Possible Datarate: ECU RAM for Application ECU RAM for Measurement DAQ Adress Network Interfaces; > XCPonCAN ~10 kb/s > Private CAN for MC ~50 kb/s > XCPonFlexRay ~ 100 kb/s > XCPonEth. ~ 2000 kb/s Debug Interfaces: > JTAG ~1000 kb/s > DAP/LFast ~3000 kb/s > DAP2/LFast max. ~10000 kb/s CANape Plug On Device max. 8,5 m XCP on Ethernet 6
High speed ECU (RAM) measurement Data Trace Approach VX1132 Requires special Aurora µc connector ECU 6,25 GB/s AURORA 25 GB/s AURORA VX145x POD 1,8 GB/s HSSL Cable 5 GB/s HSSL2 Cable VX113x Base Module 1 x 50 MB/s XCPonEth 2 x 100 MB/s MC Tool CANape VX1135 New Requirement: Next. Gen µc: i.e 4 x 6,25 GBit/s Radar Raw Data : 100 MByte/s + XCP Data : 70 MByte/s 7
High speed ECU (RAM) measurement NXP / Renesas Radar Setup: XCP-Data + Radar-Raw Data Vector approach for Radar µc from NXP and Renesas Radar raw data must share the Aurora interface with RAM data CPU (RAM data) can be stalled in case of overload, radar raw data not» Aurora Interface overload must be avoided in any situation Radar Front-End Aurora Lanes: Radar µc 4 x 1,25 GBit NXP RaceRunner 4 x 1,2 GBit NXP RaceRunner Ultra 2 x 3,125 GBit Renesas RH850 V1 8
High speed ECU (RAM) measurement Renesas/ NXP Timing Diagram : Radar Raw Data / FFT / Object List Timing.. (50µs * 256 ) = 12,8 ms 37,2 ms = ( 50-12,8) ms Chirps Buffer in VX1000 BaseModule CHIRP Datarate 80 MByte/s (4 kbyte/50 us) over 12,8 ms Average Chirp rate = 20 MByte/s Chirp1 Chirp2 Chirp3.. Chirp 256 Pause: No Chirps received 2.nd FFT, Classification, Detection, Tracking Calcuation Mem-Sync function Copy over Trace for Objectlist 40 µs 10 µs 40 µs FFT Buffer=1 MByte/50 ms Objectlist 100 kbyte Copy Object list via Mem-Sync Objectlist 100 kbyte Objectlist 100 kbyte Objectlist = 100 kbyte / 50 ms = 2 MByte/s time 9
High speed ECU (RAM) measurement Infineon Aurix Radar Setup: XCP-Data + Radar-Raw Data Radar ECU CPU0 CPU1 2,5 GBit AURORA VX1438 POD 2 x 2,5 GBit HSSL Cable VX1135 Base Module 1or2 x 1 GBit Eth. MC Tool CANape CPU2 8 x 400 MBit Raw Data Port xxx: XCPonEth. Port yyy: RadarRaw Data protocol 10
High speed ECU (RAM) measurement Infineon Aurix Core Trace Limitation For some Aurix µc there are too less CPU Master Trace units» In consequence 1 Core must be measured via RAM Copy ( CPU load + additional RAM) Example setup for Aurix TC29x ( 3 Core): M1,M2 => Master Trace Units ( Data Trace measurement ) S1,S2 => Slave Trace Units ( Data Trace measurement ) O = OLDA ( RAM Copy measurement => This creates CPU load ) S1 M1 M1 M2 O D/PSPR0 M1 CPU0 D/PSPR1 M2 CPU1 D/PSPR2 O CPU2 11 S2 LMU SRAM, EMEM O VX1000 MEM Sync function to save RAM for measurement
Radar Raw Data Replay Every traffic situation is unique, i.e depends also on other vehicle position Modified algorithm must be validate with exact the same traffic situation Existing Algorithm in C/C++/C# could be re-used / wrapped in a vadas developer component 12
ADAS Logging ADAS Logging Setup: 500 MByte/s => ~16 TByte/Day XCPonEth. USB CANape XCP Recorder Visualisierung PC1 50 MB/s Stereo-Camara Time Sync Trigger Handling ESP-ECU Video Recorder 1 150 MB/s XCP + Raw Data Recorder 2 100 MB/s Front Radar 2 x 1GBit/s Eth. Ethernet Sensor- Fusion ECU XCP Recorder 3 XCP Recorder 4 PC2 100 MB/s 4 x 40 MB/s 13 4 x Corner Radar XCP Recorder 7
ADAS Logging Data Recording and Data Transport i.e 16 TByte Data Shock Proof Logging on SSD Li-Ion UPS i.e. EV/HEV Transfer via WLAN: ~2 weeks Transport Storage Bay Thunderbolt/USB HDD Storage Thunderbolt Docking Station 10 GBit/Eth USB.30 14 Storage Server
ADAS Logging Operational Concept: ADAS visualization Flexible and easy to use Object Editor Fast camera calibration Via Chessboard and Image processing 15
ADAS Logging Operational Concept: Taxi Driver Mode Visual feedback and performing actions Displaying information like Status of the system, few signal values, Doing actions on the touch screen like fire a trigger, execute scripts, USB LCD Touch Screen User-defined GUI USB bus powered High End Logging System 16
ADAS Logging Example: ADAS Test-Setup with several Radar + Vehicle-Cam 675 MByte/s PC CPU Load: ~50 % 17
For more information about Vector and our products please visit www.vector.com Author: Alfred Kless Vector Germany 18 2015. Vector Informatik GmbH. All rights reserved. Any distribution or copying is subject to prior written approval by Vector. V1.01 2015-12-03