ZettaScale’s Journey and Impact in Open-Source Auto Software Development

A Cloud-to-Device Platform that can run on any combination of public and private infrastructure spanning from a data-centre to a micro-controller. It supports completely decentralised communication, geo-distributed storages and queries, as well as geo-distributed computations. During an interview with Vidushi, Angelo Corsaro, CEO/CTO, ZettaScale highlighted the burgeoning shift towards open-source solutions in the automotive industry, exemplified by ZettaScale’s innovative approach with Zenoh and their efforts to overcome certification challenges.

What specific limitations of closed standards like AUTOSAR have you observed that are prompting automakers to explore open-source alternatives?

AUTOSAR Classic has some well recognised limitations when it comes to supporting more dynamic workloads that are typical of ADAS. AUTOSAR Adaptive was designed with the intent of overcoming these limitations, yet it has proven to be extremely complex, heavyweight, hard to dimension and hard to cast down to deterministic behaviours. As such OEMs are looking for alternative solutions and many of them are looking at establishing Open Source platforms that can facilitate innovations, mutualise investments and help modernise and bring more agility to the automotive industry.

In Europe one of the most important initiatives in this domain is represented by the Eclipse SDV Working group. General Motors recently released its uProtocol as a foundation to the next generation vehicle platform. uProtocol has adopted Eclipse Zenoh as the underlying communication protocol.

Another stack that is seeing some traction in Open Source is the Robot Operating System 2 (ROS 2). This stack was historically based on the OMG DDS communication protocol, but recently looked into alternatives to better support mobility. The results of an in depth study were shared in September 2023 that identified Eclipse Zenoh as the best positioned protocol to address the needs of next generation robotics platforms.

In regard to the applicability of ROS2 to the automotive industry, as of today that is somewhat limited by the lack of certification for a full ROS2 stack. Some vendors provide certification for communication abstractions provided by ROS2 but those don’t add a whole lot on the underlying communication protocol. The biggest value of ROS2 is in the library available in the ecosystem.

Finally, some OEMs are looking to adopt the OMG DDS standard or Eclipse Zenoh as the backbone of their automotive architectures.

How does open-sourcing auto software development accelerate innovation cycles and enable the introduction of new features in vehicles?

Continuous integration, continuous delivery and the adoption of more dynamic communication frameworks are the key elements that are unlocking some of the innovation potential by reducing the time to contribute and test innovations.

What hurdles do you perceive for broader adoption of open source practices across the automotive industry, and how is ZettaScale addressing these challenges?

The biggest challenge is certification of Open Source. This is more a cultural challenge rather than a technical challenge. At ZettaScale we have always embraced the best Open Source practices and have been working for the past 18 months to come up with a process that allows us to certify our software at the highest automotive safety standard, namely ISO-26262.

In what ways do you anticipate an open-source revolution in auto software impacting adjacent sectors such as self-driving technology and EV charging infrastructure?

The big potential for convergence and co-innovation is between robotics and automotive. The challenge today are certifications requirements, but there are some convergences also on this domain as more and more robotic platforms will require certification.

Could you share the origin story of ZettaScale’s Zenoh and explain why automakers are gravitating towards it as an open-source option?

As a team we have been involved in the design of some of the earliest Smart Cities in the world, since 2012. These were the first examples of systems spanning from the data-centre down to the microcontroller. Building these systems was painful. It was all about stitching together technologies to make data flow end-to-end, integrate other technologies to deal with distributed computations and then struggling with data storage. Since ideally, we would have loved to keep some data stored on the edges of the system, but that would quickly complicate the retrieval.

Back then it was apparent to me that existing technologies were not designed to deal with these kinds of problems. They were not designed to be able to run and perform from extremely constrained environments up to the data-centre. Additionally, it was commonly accepted that it was normal to use one technology stack to move data, another to store and retrieve it, and yet another to deal with computations. All of this while dismissing the importance of location transparency.

I was literally obsessed with the idea of finding a way to unify all of this. Looking for a “theory of everything” that would allow me to deal with data in motion, data at rest and computation in a simple and elegant manner.
Thus, I decided to take the challenge. After a couple of years of hard work and experimenting, the team and I started developing something that was heading in the right direction. Eventually that became Zenoh, the Pub/Sub/Query protocol that unifies data at rest, data in motion and computation from the microcontroller up to the data-centre. A protocol that has no topological constraints and only 5 bytes of minimal wire-overhead!

How does ZettaScale envision its role evolving amid this industry transition towards open-source auto software development?

We have an active role as members of Eclipse. In addition to Open Source, we are working to bring Rust into automotive.

What unique relationships or partnerships has ZettaScale formed with automakers to support their adoption of open-source solutions?

We work closely with GM, BOSCH, ETAS, and many other automotive giants in the context of the Eclipse SDV to advance the state of the art of automotive software and open source.

Can you elaborate on any specific initiatives or projects ZettaScale is undertaking to further promote open-source collaboration within the automotive sector?

Besides the initiative mentioned above, we have recently started a new project called Zenoh-Flow. It provides a data-flow programming framework that will help express the typical ADAS and autonomous vehicle pipelines.