ANDATA News

The hard things about hard ODD and system specification in Vehicle Safety and Automated Connected Driving

ODD (Operational Design Domain) is a commonly used term in the context of Cooperative, Connected, Automated Driving (CCAD), Integral Vehicle Safety and Traffic Automation.

In any case, we are convinced that the methodological concepts behind that term could reinforce their significance for development, validation, and assessment of automated and connected driving functions and vehicle safety systems even more, if ODDs would be treated in a some more formalistic and harder manner. By further sharpening the definitions and establishing a precise and common view within the community, the ODD methodology could become a fundamental issue in making AI based machines and systems much more explainable and trustworthy. We have summarized the arguments for this in the preprint paper at http://dx.doi.org/10.36227/techrxiv.170654449.94227554/v1 or at ODD-Paper@Researchfate.

Why it would be worth to take a deeper look into our survey paper:

  • Formalistic, "hard and crisp" ODDs are one (of several) pillars to make complex systems and AI solutions explainable and as consequence trustworthy.
  • Software-defined vehicles (SDV) are increasingly entailed by interconnectivity between different functions. ODD can become a major instrument for the management of the complexity of this interconnectivity. This is valid internally for connections between the inner vehicles functions as well as externally for the connectivity of the vehicles with other vehicles, with the infrastructure, and with other external services.
  • Managed ODDs can become a major solution component for assurance of functions safety of C-ITS Day2+ services with bilateral and multilateral interactions.
  • Clear, transparent and committed ODDs are a cornerstone of safe cooperative, connected, automated mobility services. Knowing (or precisely estimating) others ODDs, capabilities and intentions is essential for effective cooperation and cooperative services.
  • Collective Perception does not become effective for cooperative system-of-systems without also sharing intentions and capabilities. ODD are part of that.
  • ODD is a significant part for explainability and expression of trustworthiness in complex system-of-systems.
  • The ODD methodology is a need to make effectiveness and impact assessments (like P.E.A.R.S., ISO 21934) comparable.
  • A clear and crisp understanding of the ODD is an absolute must in scenario based development, assessment and validation of automated and connected vehicles.

Hence a transparent, clear, tangible, machine understandable (and not only readable!) formulation of ODDs is relevant for numerous topics, initiatives, and standards, like

  • prospective effectiveness assessment (cf. P.E.A.R.S., ISO 21934, etc.)
  • OpenODD from ASAM e.V.
  • test scenarios for automated driving systems (cf. ISO 34505, OpenScenario, etc.)
  • trustworthiness of AI-based systems (cf. ISO IEC/JTC1/SC42 WG 3)
  • Car 2 Car Communication Consortium with C-ITS usecases Day 2+ as well as 5G Automotive Association (5GAA)
  • Infrastructure Support for Automated Driving (ISAD)
  • CCAM in general

Also be aware that the paper and the underlying approach is relevant in the same way for autonomous ships as well (> IMO).

Please feel free to comment and discuss with us in Linkedin or directly!

Inclusion of Expectator and the FEM Operations Toolbox in the Mathworks 3rd party connections programm

We really appreciate also being included in the Mathworks 3rd party connections program with our products Expectator and the new FEM Operations Toolbox!

20 years ago, we made the deliberate and well-considered decision to build  our AI solutions on top of the trustworthy and high-performance mathematics of Mathworks.

With the new FEM Operations Toolbox, we have once again been able to do things that are not available anywhere on the market in this form. That way we strive do enable absolutely new approaches in the development and approval of safety critical applications, for example like in integral automotive safety.

The FEM Operations Toolbox and Expectator are the fundamental parts for solutions, which ANDATA collocates under the brand of commodo (complex modelling and simulation operations), which combine the Finite Element Method (FEM) and Computer Aided Engineering (CAE) with Artificial Intelligence (AI).

Thank you Mathworks for the steady support and responsiveness. We keep on pushing on the safe way to VisionZero! Expect more to follow…

CAE and AI with commodo and the FEM Operations Toolbox at NAFEMS

For more than 15 years now, ANDATA has been working on the combined application of Artificial Intelligence (AI) methods in Computer Aided Engineering (CAE) under the title commodo (complex modelling and simulation operations, see failure prediction). Long a niche topic for experts and innovators, it is now on the threshold of entering the mainstream.

With the FEM Operations Toolbox, ANDATA is now providing an enabler for the direct application of AI and machine learning methods on and in finite element simulations, in particular crash simulation.

We present field reports and application examples at the NAFEMS conference on October 24th, 2023 in Munich.

We are firmly convinced that the combination of CAE with AI, among other things, is a major milestone in the development of future vehicle safety systems on the way to VisionZero. And this applies both to

  • improved and realistic simulation in product development as well as
  • in primary simulation-based system validation and product approval.

For more information, come to the NAFEMS event on October 24th or contact us directly.

Robotaxis producing traffic chaos in San Francisco

According to recent news like CNN: ‘Complete meltdown’: Driverless cars in San Francisco stall causing a traffic jam, Robotaxis from Cruise and Waymo are producing great chaos in the streets of San Francisco.

Unfortunately, our mathematically thorough predictions from several years ago seem to become true.

🚦 These traffic problems of autonomous vehicles have already been identified and named at an early stage within the WienZWA project (see Will autonomous cars solve our traffic problems?, Balancing the benefits of automated driving, and Test Fields and Advanced Accompanying Methods as Necessity for the Validation of Automated Driving).

💡 The reasons are obvious and clearly described. Of course there may still be hard technical challenges at the level of sensors and control algorithms. The geeks from Cruise and Waymo will solve them sooner or later by clever, neat, and systematic engineering. But there are still profound conceptional burdens at the level of traffic and the interactive behaviors of the according systems and divers traffic participants. Such cannot be solved by looking at the vehicle's technology only. Their resolution requires a broader community approach and contribution of other stakeholders too.

Fortunately, concrete solution strategies are known and already in the making respectively available with higher degrees of maturity:

👉 Infrastructure Support for Automated Driving (ISAD) to

  • deliver actual information from original sources for valid driving decisions,
  • expand vehicle’s ODDs with external help,
  • provide dynamic risk rated maps for congestion/accident prevention,

👉 Virtual Traffic Control to

  • coordinate automated vehicles in a self-organized manner,
  • enable active cooperation,

👉 Automated, self-organized, distributed Traffic Management & Control to

  • sort out incidents and congestions quickly,
  • select the best strategies upon the given/predicted situations,

👉 Digital Twins of the traffic infrastructure for

  • multilateral and trustworthy exchange of shared information between infrastructure and traffic participants,
  • consolidation of shared information from different sources with different levels of trust,

👉 Distinct and holistic CCAM System-of-Systems Architectures to

  • bundle the above items in a controlled and orchestrated way,
  • reduce the overall complexity in traffic.

More details on how to sustainably overcome the encountered challenges and how to realize the benefits of automated driving nevertheless, can be found under the umbrella of VERONET.

Also watch out here for more upcoming initiatives soon (>FUNDAMENTO, DESIRE2, DETECTION, NORBERT, ODD, ...).

FISITA World Congress 2023

Before leaving for summer break, we want to make a short announcement:

The FISITA World Congress 2023 will take place in Barcelona beginning on September 12th.

ANDATA will immediately open the first technical session on "Integrated safety, connected & automated driving" with a presentation on "Criticality estimation for connected, cooperative collision avoidance".

Ex ante some key issues:

👉 Within the project DIGEST we began to define fundaments for Digital Twins of the traffic infrastructure, containing information for improved driving decision in Cooperated, connected, Automated Driving, which we call the "CCAM Decision Support Platform".

👉 Cornerstone for information exchange and cooperation is the definition of a common language to express observations, intentions, capabilities, and ODDs.

👉 The project COPE (Collective Perception) showed how such information may look in a generic way to avoid collisions between any kind of traffic participants, especially for vulnerable road user protection.

👉 Within the project IntIntSec we will incorporate such into a holistic Digital Twin of an intelligent intersection, finding the best compromise between vehicle efficiency (saving energy), traffic efficiency (avoiding congestions), and safety (staying collision free).

We deliver the control algorithms and the underlying technological framework.

Eager to see you in Barcelona, paving the safe way to Vision Zero.

 

Further details can be requested from info@andata.at.