Welcome to the Drive World with ESC 2019 Presentation Store. Here you can view and download conference presentations before, during, and after the event. If you’re looking for a presentation from a specific session that you’re unable to find here, note that it’s likely because the presenter has not provided permission for external use or has not yet shared their presentation with us. Please check back after the event for a more complete catalogue of available presentations.
Car 2.0: Transportation-as-a-Service
Willard Tu (Senior Director, Automotive, Xilinx)
Location: Room 206
Date: Thursday, August 29
Time: 11:00am - 11:45am
Track: Drive - Sensing Technologies
Format: Technical Session
Vault Recording: TBD
The march to automated driving is quickening with the infusion of investment capital, innovation in artificial intelligence, and public expectations. Automotive innovators are demanding adaptable and scalable solutions to achieve Transportation-as-a-Service nirvana. However, each has a different approach to solving the problem that span:
- Artificial Intelligence: Computer Vision vs. Neural Nets
- Computer Vision – demands more performance, which drives cost up and thermal dissipation, better suited for FuSa
- Neural Nets – are a true black box, but will likely be lower performance than Computer Vision and lower thermal
- Compute: Distributed vs. Centralized
- Most of the traditional passenger owned vehicles are driven by costs. These vehicles today are distributed intelligence so that OEMs can source an ECU = 1 function to a supplier. Centralization is much more complex as either the OEM or a Tier 1 has to take on greater responsibility for multiple functions that might be put into a centralized computing center.
- Centralization has trade-offs — you now have to stream a lot of data to a central node this is not easy. Cost is moved from processing at the edge to data transportation across the vehicle
- Robotaxi — vendors are doing it completely differently. They are all for centralization, cost as strong a consideration.
- Sensing: Camera, Radar, LiDAR
- We will discuss the trade-offs and likely cost projections of each technology.
- Processing Engines: CPU, DSP, FPGA, GPU
- Will contrast and compare each of these engines.
- Latency: Batch vs. Batch-less will be covered in this comparison.
In this session, Willard Tu, senior director of automotive for Xilinx, will highlight the innovator profiles, the constraints that they are facing, the market forces influencing them, and the ecosystem that strives to help them.