accele lerating av
play

Accele lerating AV Productization wit ith AI Danny Atsmon - CEO, - PowerPoint PPT Presentation

Accele lerating AV Productization wit ith AI Danny Atsmon - CEO, Cognata Simon Berard CATIA Strategy Senior Manager, Dassault Systemes Agenda Vision Challenges AI Solutions in Design AI Solutions in Validation


  1. Accele lerating AV Productization wit ith AI Danny Atsmon - CEO, Cognata Simon Berard – CATIA Strategy Senior Manager, Dassault Systemes

  2. Agenda Vision ● Challenges ● ○ AI Solutions in Design ○ AI Solutions in Validation Solution Convergence ●

  3. AI AI-driven Desig ign and Vali lidation Solu lutions

  4. Vis ision

  5. Design Challenges

  6. Chall llenge: Mis issio ion Driv iven Engin ineerin ing Mission lef eft t to o the Use User Mission en engineered within the Sys yste tem 3x more System Autonomous Traditional Vehicles Systems Requirements People Delivery Taxi Driver

  7. Costs vs Comple lexity – AI I is is makin ing it it possible MOBILITY EXPERIENCE __________ AUTONOMOUS Business Driven VEHICLES Innovation __________ Leverage Patrimony MASS- CUSTOMIZE ZED CARS RS __________ Explore Variants COMPONENTS __________ Structure Legacy Functional Requirements

  8. AI I is is brin ingin ing Quali lity Functional Requirements

  9. Chall llenge 2: : Physic ically Exa xact Multidiscipline, Multiphysics, Multiscale Sensors optimization Consistent System Experience Validation

  10. AI I Solu lutions fo for Design Function Driven Learning from Patrimony Generative Design Context Sensitive Model Based System Automated Assembly Engineering Parameters space Performance Exploration Tradeoff

  11. Valid lidation Challenges

  12. AV Tech Has Had Some Unpla lanned Setb tbacks

  13. Chall llenge #1: Scale L4: DR DRIVE __________ Driverless Taxi L2+: ASSIST __________ Highway L2: ASSIST __________ L1: ASSIST Chauffeur Level of Autonomy __________ Adaptive L0: Cruise INFO FORM Adaptive Control __________ Cruise + Blind Spot Control Lane Lane Centering Warning Park Assist Functional Requirements

  14. Chall llenge #1: Scale - Need Vs. . Actu tual The Need: 11 11B* B* Actually dri riven: 16M (0 (0.15%) *Rand corporation

  15. Chall llenge #2: Realis ism REAL LIF IFE SIM IMULATION VS. Realism Scalable Realism Scalable 100% ??% We need a reali listic sim imulation fo for a meaningful l coverage

  16. Reali lism - Uncanny Vall lley (M (Masahiro Mori, i, 1978) Moving Fully human + UNCANNY Still VALLEY Humanoid robot EMOTIONAL RESPONSE Bunarkupuppet Familiarity Stuffed animal Industrial robot Polar Express Human likeness 50% 100% Completely machine - like Prosthetic hand Zombie Simulation today 16

  17. Reali lism - Uncanny Vall lley Moving Fully human + UNCANNY Still VALLEY Humanoid robot EMOTIONAL RESPONSE Bunarkupuppet Simulation Familiarity needs to be here Stuffed animal Industrial robot Polar Express Human likeness 50 50% 100% Completely machine - like Prosthetic hand Zombie 17

  18. The anim imation solu lution “ An animator cannot capture all of reality . Instead he picks 3 or 4 distinct elements and exaggerate them. “ - Walt Disney 18

  19. AI I based meth thods fo for Realis ism Procedural Modeling of a Building from a Learning from Synthetic Humans Nvidia & MIT - Video (Labels) to Video Single Image – Nishida et al, 2018 – Varol et. al 2017 Synthesis – Wang et. al. 2018 END TO END LAYERED APPROACH LAYERED APPROACH Realistic, not consistent Consistent, Not scalable Consistent, Not scalable (Manual) (Variations)

  20. Chall llenge #2: : Realis ism conclu lusio ions REAL LIFE SIMULATION VS. • DNN transfer functions ~= Realism • Isolated layers brings better results than End to End Data sources should be wide (crowd sourced) •

  21. Cognata - 4 4 Technolo logy lay layers STATIC DYNAMIC SENSING CLOUD & ANALYTICS 21

  22. A Combined Solution

  23. Desig ign and Vali lidation Converge Better Tog ogether ● Tests and designs together ● Validation base directly from requirements ● Smart coverage

  24. Takeaways Autonomous vehicles brings ● ○ New designs and use cases Large scale validation challenge ○ AI is a key to get Autonomous vehicles in a safe and cost effective way ● ● A platform solution to manage, design and validate is the needed solution

  25. Thank You!

  26. End to to end li limitations The Problem: Not consistent, overfeat 26

  27. Buil ildin ings reconstruction conclusions The Good: Consistent, Procedural The Bad: Not practical 27

  28. Learning pose - Conclu lusio ions This the most advanced way of learning moving objects. The Good: Consistent The Bad: Not scalable 28

Recommend


More recommend