Apollo Cyber RT — The Runtime Framework You’ve Been Waiting For
Learn more about why Apollo Cyber RT is your go-to runtime framework to build your own autonomous driving solution.
The Apollo Cyber RT framework, launched as part of the Apollo 3.5 release, marks a new and significant milestone in the evolution of the Apollo platform. It is the world’s first open-source, high-performance runtime framework designed specifically for the development of autonomous driving technology. What can it do, you ask? Let’s dive in.
The goal we set for the Apollo Cyber RT framework was simple: to give you a robust framework with high performance computing to support and manage the complex tasks of urban driving. It can effectively handle the unexpected situations and challenging scenarios that self-driving technology can run into. This all starts with an efficient design.
At the basis of the architecture, you’ll find a set of components with predefined inputs and outputs. Each component contains a specific algorithm module built to process a set of data inputs and generate a set of outputs. The Apollo Cyber RT framework is built on top of these components. It extracts the dependencies from the components and links all of them together using a DAG dependency graph.
At runtime, the framework takes the predefined components described above and combines them with fused data from the sensors to create lightweight user-level tasks. Each task is then scheduled based on resource availability and task priorities and executed as optimized threads.
Accelerate Development
The design of the Apollo Cyber RT framework offers a centralized and parallel computing model which enables high concurrency of task execution, low latency and high throughput. The adaptive communication capabilities of the framework allows for high data throughput while the user-level scheduler with resource awareness enables the centralized computing model to perform at the highest level. That means you get optimized data processing and communication out-of-the-box when you use the Apollo Cyber RT framework to build your self-driving application. We’ve already integrated a well-defined task interface and efficient data fusion to let you move directly to building your top-layer solutions without having to worry about performance or efficiency of the system.
The Apollo Cyber RT framework also gives you access to a vast array of development tools like Cyber Monitor, Cyber Visualizer, Cyber Recorder and a suite of pre-integrated sensor drivers. In other words, we’ve done our best to accelerate the development of your autonomous driving system without compromising performance or brand trust.
Simplify Deployment
We wanted the Apollo Cyber RT framework to be fast, but we also wanted it to be flexible to simplify deployments. That is why we made it a stand-alone framework with minimum dependencies. We wanted you to have the ability to take our framework and integrate it into your self-driving system the way you see fit. We also made the user scheduler highly configurable to adapt to any type of self-driving application and provide you with the highest level of flexibility.
But simplifying deployment also means not having to reinvent the wheel. So we created a set of standardized out-of-the-box top-level modules that you can add to your system to get your solution off-the-ground in no time. These building blocks include Planning, Control, Perception, Localization and Human-Machine Interface (HMI), and they all integrate seamlessly on top of the Apollo Cyber RT framework.
Building the Future of Autonomous Driving with You
But there is more. Integrating with our own AV modules is one thing, but we wanted our framework to be highly versatile. The Apollo Cyber RT has a plug-and-play architecture that can integrate not just with our building blocks, but with any AV software modules. This gives you the ability to integrate your own AV modules to work on top of the Apollo Cyber RT framework, or in combination with the Apollo building blocks. This unique design empowers developers like you to build your own autonomous driving solutions quickly and efficiently and the best part is that it is completely open sourced.
To help you visually learn more about Apollo Cyber RT framework, here is a animation video we prepared for you:
Access the new Apollo 3.5 code on our GitHub page now!
Our mission to accelerate the deployment of safe and reliable autonomous vehicles worldwide starts with you. Autonomous driving solutions built on the Apollo Cyber RT framework will benefit from the highest level of performance without requiring heavy and complicated deployment. As such, we hope it will encourage thousands more developers to join the Apollo platform and create their very own autonomous driving solution.
And stay posted: Follow us on Twitter @ApolloPlatform and sign up for our newsletter here.
A special thanks to Natasha Dsouza, Ning Qu and Apollo’s CarOS team who helped with technical content of this blog post!
