Apex.AI

Apex.Grace keeps the ROS 2 publish and subscribe model, the request and reply pattern and the node lifecycle that most teams already know. Extensions cover real-time execution, system state management and fleet deployment. In practice your prototype code survives the jump to production, which is the whole reason teams pick this stack over AUTOSAR Adaptive.

The runtime is certified to ASIL D, the strictest level in the automotive functional safety standard, and the company states the assessment was done by TÜV NORD. That certificate is what lets an OEM put the software on a safety-critical path without rebuilding the framework from scratch. You still certify your own application, but you start from a qualified base.

The executor lets integrators group work into single items, sequential chains or directed acyclic graphs, then run them on a thread pool. Thread priorities, core affinities and scheduling policies are all configurable. Separating application logic from execution policy means the same code can meet a hard real-time budget on one ECU and a softer one on another.

Apex.Ida routes messages through shared memory and avoids the copies and context switches that eat latency in a normal DDS setup. Apex.AI reports throughput in the hundreds of gigabits per second for embedded targets, with low jitter under load. For a perception pipeline pushing multi-megabyte camera frames, that difference shows up as stable timing instead of spikes.

A single application-facing API sits in front of DDS, SOME/IP, MQTT, CAN and WebRTC. The middleware picks the transport from configuration, so a radar signal from CAN and a camera frame from shared memory look the same to your code. This is what makes the automotive middleware usable in a vehicle that mixes old and new networks.

Apex.Autonomy is an autonomy stack built on Apex.OS for UGVs. It covers autonomous navigation, follow-me mode, convoy operations, evacuation routing, and object and drone detection. Because it runs on the same foundation, autonomy features can be updated without replacing the vehicle platform.

Apex.Alan handles integration, build, test and delivery pipelines, and the V&V package records and replays very large data streams with time alignment. Teams use it to reproduce field failures on a bench. The tooling matters as much as the runtime, because safety evidence is mostly a documentation problem.