Real-time computing and platforms vs. Smart algorithms and application of tomorrow

Embedded systems are evolving at a rapid pace, with new challenging applications requirements, contexts and algorithms, as well as changing computing architectures. The workshop has the goal to discuss how predictability and other important properties can be achieved for future automated and smart real-time systems.

Automated driving can be taken as a highly representative example. Application level tasks involve perception, localization, planning, decision making and control in highly complex environments.
The types of algorithms range from rule based, and traditional control systems, to sophisticated (and sometimes non-deterministic) planning as well as machine-learning systems. Both the hardware and software platforms are evolving towards highly heterogeneous and complex systems in themselves, featuring for example Linux, middleware, multicore and GPU’s, as well as traditional microcontrollers and RTOS’es.

What are the characteristics of applications for future automated and smart real-time systems?
How can the platforms support dependable computing for these types of algorithms?
What can be learnt or reused from existing computational models such as those developing in scheduling theory (e.g. anytime/imprecise computations) and models of computation?

Workshop format:
-    Half day at KTH: 09.30 – 13.00
-    Starting with registration and coffee served from 09.00

Workshop programe:  
- Workshop Introduction – contrasting algoritms, architectures and models of computation/patterns – where are we? (Martin Törngren)
- Developing Predictable Vehicle Software: The Rubus Approach - Saad Mubeen
- Correct-by-construction-design - Ingo Sander, KTH and Ingemar Söderqvist, Saab
- Lars Svensson and Xinhai Zhang, KTH - "new" automated driving functionalities and their requirements
- Linux och hypervisors Ola Redell, Retotech
-    Lunch wrap served for those registered at 12.00