Hamid Tabani, a research engineer at Barcelona Supercomputing Center (BSC), made a presentation titled "Challenges and Opportunities Towards Full Autonomous Driving" at the HiPEAC Computing Systems on Wednesday, 14 October 2020 at 14.00h that explains part of MASTECS' technology.

The presentation explained both challenges and opportunities that lie ahead on the road to full autonomous driving. It introduces the audience to autonomous driving systems and the challenges they brought to the real-time, automotive, and computer architecture communities. The program was as follows:

  1. An Introduction to Autonomous Driving Systems: An Industry-level Case Study.

  2. Safety Standards and their Certification Requirements.

  3. Existing Challenges toward Certification of Autonomous Driving Systems.

  4. Autonomous Driving Systems: The Extensive Use of Deep Learning.

  5. Solutions toward Accelerating Compute-Intensive Modules in Autonomous Driving Systems.

  6. Timing Analysis and Resource Usage of Advanced Autonomous Driving Systems.

Overall a range of ideas were explored in the presentation: At first, an introduction to basic concepts of industry-level autonomous driving systems is provided. Secondly, an analysis of the main requirements from performance guarantees to safety is covered. Thirdly, an analysis of the main approaches for hardware design to fulfill those requirements with an emphasis on the work done as part of the European Processor Initiative (EPI) was presented. Fourthly, the presentation focused on the deep learning evolution and their wide usage in autonomous vehicles, timing, and safety-related aspects for them. Finally, an industry-level autonomous driving software stack and its modules as a representative cyber-physical system case study was discussed.

Watch the full presentation here.