DATE is pleased to present a special hybrid format for its 2022 event, as the situation related to COVID-19 is improving but safety measures and restrictions will remain uncertain for the upcoming months across Europe and worldwide. In transition towards a future post-pandemic event again, DATE 2022 will host a two-day live event in presence in the city of Antwerp (just north of Brussels in Belgium), to bring the community together again, followed by other activities carried out entirely online in the subsequent days. This setup combines the in-presence experience with the opportunities of on-line activities, fostering the networking and social interactions around an interesting program of selected talks and panels on emerging topics to complement the traditional DATE high-quality scientific, technical and educational activities.

ASDW05-05 Predictable Perception

Session Start
Fri, 14:00
Session End
Fri, 14:55

 

  • Organizers / Chairs: 

    • Samarjit Chakraborty (U North Carolina, Chapel Hill, USA)

    • Petru Eles (Linköping University, SE)

  • Session Abstract:

    Modern autonomous systems - such as autonomous vehicles or robots - consist of two major components: (a) the decision making unit, which is often made up of one or more feedback control loops, and (b) a perception unit that feeds the environmental state to the control unit and is made up of camera, radar and lidar sensors and their associated processing algorithms and infrastructure. While there has been a lot of work on the formal verification of the decision making (or the control) unit, the ultimate correctness of the autonomous system also heavily relies on the behavior of the perception unit. The verification of the correctness of the perception unit is however significantly more challenging and not much progress has been made here. This is because the algorithms used by perception units now increasingly rely on machine learning techniques (like deep neural networks) that run on a complex hardware made up CPU+accelerator platforms. The accelerators are made up of GPUs, TPUs and FPGAs. This combination of algorithmic + implementation platform complexity and heterogeneity currently makes it very difficult to provide either functional or timing correctness guarantees of the perception unit, while both of these guarantees are needed to ensure the correct functioning of the control loop and the overall autonomous system. This is a part of the overall challenge of verifying the correctness of autonomous systems.

  • Speakers:

    • Qing Rao (BMW, Munich, Germany) 

      New Era in Autonomous Driving and the Role of IT - Will Traditional Carmakers Keep Pace?

    • Soheil Samii (General Motors R&D, USA) 

      Dependable sensing system architecture for predictable perception in autonomous vehicles

    • Deepak Shankar (Mirabilis Design, USA) 

      Design Tools for Predictable Hw/Sw Architectures for Autonomous Vehicles

    • Cong Liu (UT Dallas, USA)  

      Towards Timing-Predictable & Robust Autonomy in Autonomous Embedded Systems

    • Hamed Tabkhi (University of North Carolina at Charlotte, USA) 

      Toward AI-in-the-Loop Autonomous Safety System - Algorithmic and Timing Challenges