As mobility shifts to a more electric, connected, and autonomous future, the way vehicles are architected is also changing. Software-defined vehicles are the way forward, say most experts.

One company leading the charge is Apex.AI. The Palo Alto-, Berlin-, and Munich-based developer of next-generation scalable software for mobility systems was cofounded in 2017 by Jan Becker, President and CEO, and Dejan Pangercic, CTO. Helping them innovate at the software-defining company are Paul Balciunas, CFO; Sanjay Krishnan, VP of Product; Serkan Arslan, VP of Business Development; Tavis Szeto, VP of APAC; Michael Pöhnl, Senior Staff Engineer; and Zion Maffeo, VP of Legal and Business.

The company’s flagship product is Apex.OS, a meta-operating system that enables faster and easier software development for vehicles and safer autonomous driving solutions. Its software development kit targets two main customer groups, the OEM developers who need to implement complex AI (artificial intelligence) software integrated with the entire vehicle, and the autonomous driving developers who can now implement safety-critical applications with ease and reliability.

From the beginning, the company says its Apex.OS has been engineered to be a customer-oriented and developer-friendly framework for creating modern, automotive-oriented software. The German Design Council recognized the company’s work with two awards for the automotive end-to-end operating system: Best of Best in the Mobility & Innovation category and Innovation of the Year.

Automakers are also recognizing the value the company adds. Toyota’s Woven Planet Group is working with Apex.AI to develop and deploy a production-ready autonomy stack for Arene, Toyota’s vehicle development platform to enable modern software development tools and best practices in the automotive industry. Woven Planet is using Apex.OS, an SDK (software development kit) for automotive and other safety-critical applications including autonomous driving software.

Announced in August, the latest update to Apex.OS, version 1.3, gets the meta-operating system for mobility, smart machines, and IoT another step closer to meeting customer needs for software-defined vehicles.

“Our customers work on applications for advanced driver assistance systems or autonomous driving with multiple sensors and software components based on Apex.OS,” said Becker. “These applications are built as complex graphs with a combination of data- and time-driven workflows that need to run deterministically and in real-time. Apex.OS 1.3 now simplifies the development of a safety-certifiable software stack because the new version has a deterministic executor.”

Apex.OS 1.3 addresses the desire to deterministically replay the data of systems under test. Determinism allows the execution of tests and validations required for the use of functions in the automotive industry on public roads and forms the basis for real-time executions.

With this capability, it is now possible to replay a simulated or recorded scenario any number of times and with the confidence that the same behavior will be observed each time it is replayed. This makes validating a software stack to meet the ISO 26262 standard much easier.

Apex.OS is ASIL-D certified, the highest requirement of ISO 26262, since the launch of Version 1.2. No comparable operating system is offered by a competitor, Becker told Futurride.

“Nobody else is certified at this point,” he said. “Also, nobody else is based on open APIs, and nobody else has a software that combines all the different transport mechanisms in one.”

Adhering to the requirements of ASIL-D was relatively easy for the software-focused company.

“It’s actually not rocket science,” said Becker. “It’s just a lot of engineering discipline and being smart. What we’ve done in one year takes the typical automotive company three to four years in the certification process. We were just very lean, very disciplined. And, in terms of software, very clean from the beginning.”

What advantage does the ASIL-D certification bring?

For example, with by-wire steering, “we can run in a system that requires ASIL-D overall certification without software redundancy,” said Becker. “So, we don’t need a second set of software. We could be the one single operating system in a steering ECU.”

The 1.3 update allows Apex.Middleware to be used in conjunction with the OS system, addressing challenges that can arise when software components run on different processors generating gigabytes of data with latencies of microseconds. The middleware guarantees real-time data transmission with “zero-copy” communication between sender and receiver, pushing latency to a minimum and reducing processor utilization.

The software is designed to be used with any other development framework including AUTOSAR Adaptive. Apex.Middleware was developed as a highly optimized integration of the Eclipse Foundation’s DDS and Iceoryx.

Documentation for Apex.OS users has been expanded significantly to facilitate developer productivity. This includes instructions for porting code from ROS or ROS 2 robot operating systems to Apex.OS and a description of how to use the zero-copy feature to ensure extremely low latency for large amounts of data such as camera and lidar images.

As Becker describes it, a camera image is typically copied for many computational processes.

“Camera images are large, and cameras take a lot of images, and you have a lot of cameras on the vehicle,” explained Becker. “So that then leads to an insane amount of data, and if you copy all of them, it takes an insane amount of time. With existing processing power, you then have a hard time doing that in real-time, you have a hard time storing in the memory, and so on.”

The Apex-OS solution gets rid of all that copying.

“Once the camera images arrive in the memory, it is never copied again,” he said. “We have developed a way how different processes can read out that one camera image and all work on the same copy without interfering with it.”

To support the work of the ROS community, Apex.AI has contributed features and performance improvements to the open-source ROS 2 project.

“With our continuing development of Apex.Middleware, we were able to make significant contributions to the Eclipse Cyclone DDS and Eclipse Iceoryx projects,” said Becker. “We are pleased to be able to advance these open-source projects and would like to thank the community for making it possible with their work to develop and deploy advanced applications without having to constantly reinvent the core building blocks.”

With the upgrade to 1.3, these core building blocks, which are firmly anchored in Apex.OS, now have an even greater chance of becoming an industry standard for the mobile world.