Keep on Hacking: Eclipse SDV , Microsoft ThreadX and more
Bring together bright minds, give them something to play with (and lots of caffeine, food, and coaching), e voilà: It's amazing what can be achieved in a 48 hours!
This week, Accenture was hosting the Eclipse Software Defined Vehicle hackathon at its Munich office. We are a regular contributor to the SDV working group of the Eclipse foundation on the base SW stack, and it was great fun to run this year's challenge. 15 teams worked on new business and tech solutions, based on the foundations of the SDV working group. Although, no wonders are to be expected in just a few hours, the teams showed great results and thought "out the box". I've seen three different angles of innovation:
new functions/elements for the tech stack (e.g. adding pre-sensing functions for improved pedestrian safety)
combination of existing elements to new business models (e.g. implementing use cases for car sharing companies)
process innovation (e.g. adding tracking and debugging intelligence)
Half of the teams came already as a team (mostly with a company background), while the other half of the teams just formed at the hackathon - amazing, how fast and well this worked! It wasn't easy to assess the teams' performance - but I had great support in the jury: Gerd Schaefer from ESR Labs AG, Andre Larberg from IAV, Boris Engel from Microsoft and Thomas Fleischmann from CARIAD.
The Lego principle
Why did this work so well, and so easily? The trick is Lego. Or, a bit more precise: having a basket of well-defined, pre-built tech blocks with standardized interfaces.
In the hackathon for example, many teams used the #eCAL framework, contributed by Continental and now part of the Eclipse library (see the github repo here):
eCAL (enhanced Communication Abstraction Layer) is a fast publish-subscribe middleware that can manage inter-process data exchange, as well as inter-host communication.
And not only the library was available at Munich, but also a real prototype vehicle to implement on. Thanks to Continental for the great support!
The key principles of the Eclipse foundation are four:
1. Projects: The working group has initiated more than 20 projects - see here:
2. Open Technology Platform: open-sourced, modular software components and frameworks, with emphasize on automation and virtualization.
3. Automotive-Grade Solutions: prioritizes high-performance computing and maintains high standards in quality management, security, and safety across all vehicle domains.
4. Open Standards: uses existing open standards wherever possible, fostering open collaboration and innovation without reinventing existing technologies
The Lego bricks available at the SDV working group were lowering the entry barrier massively, speeding up the onboarding also for teams with only little automotive experience so far. This is especially remarkable as a lot of team members are current students at the 42 school, which educates students outside the linear & standard CV with a pragmatic and application focused approach, (greetings to Max Senges and colleagues!).
Open Source is about Contribution - ThreadX
Last week, another big contribution was made by Microsoft - open-sourcing the #ThreadX Real-Time Operating System (RTOS), which was acquired through Express Logic. This move, which sees ThreadX and the Azure RTOS development suite being donated to the Eclipse Foundation, noteworthy also for its widespread use, could bring additional dynamics into the Automotive world.
Microsoft claims that ThreadX runs on approximately 12 billion devices, including being the firmware that controls every Raspberry Pi larger than the Pi Pico. It's relevant for several reasons (read more on the spec here):
Real-Time Capabilities: RTOS like ThreadX are designed for real-time applications, which are crucial in automotive systems where timely and predictable responses to sensor data are vital for safety and efficiency.
Wide Adoption and Reliability: The extensive use of ThreadX in various devices underscores its reliability and performance, traits that are essential for automotive applications. Good starting point also for making a standard.
Community and Innovation: By making ThreadX open source, Microsoft is fostering a community-driven approach to development. This can lead to more rapid innovation and improvement, as developers and companies can contribute to and benefit from a shared technology base.
Safety is a beast
But Automotive needs also safety - which means functional safety, ISO 26262 and more. And there is still a fundamental lack of systems which comply with safety requirements (ASIL B/C/D) and SW best practice like POSIX. The "Portable Operating System Interface" is a set of standards specified by the IEEE for maintaining compatibility between operating systems. POSIX standards define an environment that can be hosted by various multi-tasking operating systems, offering a supervisor mode and a protection mechanism.
But that's not sufficient yet for being also "safe":
Safety Culture and Organizational Structure: Implementing a safety culture is critical in safety-critical product development. This includes assigning an independent Functional Safety (FuSa) manager with authority over the product development team, and adhering to a documented safety plan and safety specifications.
Safety Mechanisms in Hardware and Software: The design team must integrate functional hardware and software safety mechanisms such as Error-Correcting Code (ECC), parity checking, and dual-core lockstep systems. These mechanisms are outlined in the Safety Concept specification and are crucial in designing safety-critical System on Chips (SoCs) and IP products.
Design and Assessment of Safety Mechanisms: Hardware/software safety mechanisms must be assessed and validated against ASIL ISO 26262 safety requirements. This involves analyzing the impact of these safety mechanisms on identifying and correcting potential faults in the IP during the verification and validation phase of the SoC development process.
Random Fault Analysis: Compliance with ISO 26262 requires a random fault analysis focusing on hardware safety. This includes Failure Modes, Effects, and Diagnostic Analysis (FMEDA) and preparing a safety manual. Key results such as Single Point Fault Metric (SPFM) and Latent Fault Metric (LFM) rates are defined to meet specific ASIL levels.
Systematic Faults Assessment: Besides hardware/software safety development, there's also a need for a safety assessment for systematic faults in the development process. This encompasses all development phases, including planning, development, verification/validation, assessment, product release, and ongoing maintenance. The systematic process ensures that multiple steps and reviews are performed, including ongoing monitoring to adhere to ISO 26262 standards.
Traceability and Documentation: There's an emphasis on traceability, confirming that implementation and verification have satisfied requirements. This is crucial for safety-critical systems. Adequate documentation and supporting evidence are necessary to demonstrate that safety goals have been met.
Robust Tools and Manufacturing Oversight: During the development phase, engineers must use robust software tools to prevent introducing or overlooking design bugs. Manufacturing teams need to ensure zero defects in parts and that safety mechanisms trigger effectively in faulty conditions.
Let's act now
The challenges to make Automotive software scalable, safe and performant are big. But one key ingredient is directly in front of us: collaboration.
The number of parallel initiatives is high. Perhaps too high. All of them tackle relevant challenges, and they are contributing important building blocks. We need a better alignment of those initiatives, from Eclipse Automotive, Digital.Auto, Covesa, SOAFEE and many more.
We are working on getting to one common governance, and joining forces in a common platform that multiplies the efforts.
Let's get in touch to discuss how to reach a critical mass and speed to change the Automotive world!
#SDV #SWdriven #SWdefinedVehicle #Automotive
Juergen Reers Dennis Röhr Wolfgang Köcher Stephen (Yizhou) Xu Amarnath Bharadwaj Raul Kraus Maxim Rasovsky Bettina Blum Christian Levels Christian Kleikamp Marcello Tamietti Anurag Jain Raghavendra Kulkarni François BARTHET Hans Loes Gerd Schaefer Benno Stützel Hendrik Dettmering