Driverless car systems must be able to do three main things: perceive the environment, process the data in that environment, and act on the environment based on the information they get. In addition to the development environment, developers need to face a wide variety of internal and cloud applications, which require highly autonomous systems to translate concepts into reality.
The automotive industry has gone through a long period of time, the technology of self-driving cars and autonomous vehicles is no longer science fiction, and the future world is indeed full of expectations. The reality is much more complex, and this certainly applies to the technological environment of the future of driverless cars – especially during the proof-of-concept phase.
In addition to the unique and demanding development environment, developers struggle with a wide variety of customized on-premise and cloud applications, all of which need to be able to communicate seamlessly with each other. This is truly an Industrial Internet of Things (IIoT) system that requires a high degree of autonomy to translate concepts into reality. RTI can integrate all of this into a reliable overall operation, making engineering projects run more efficiently.
Take action
I have been with RTI for almost 4 years. In the meantime, we have witnessed more and more manufacturers throwing themselves into the torrent of driverless car development. My role is to work with sales teams, partners, and strategic customers and help them all succeed, because that's where RTI can offer a lot of services.
But when is it appropriate to ask RTI for help? When passing the proof-of-concept phase, you must circumvent the roadblocks that suddenly appear in front of you. First, driverless car systems must be able to do three main things: perceive the environment, process the data in that environment, and act on the environment based on the information they get. This is essentially a cyclical cycle, but the amount of data generated and the speed at which it needs to be processed can quickly become overwhelming.
Driverless car systems
Common challenges
Dissecting further, when we look at a driverless car, it must have a sensor package that can observe the environment, which can be either a simple assisted driving level technology or a more complex, height or fully autonomous vehicle. It will determine the level of accuracy and quantity of data collected from lidar sensors, radar sensors, actuators, and other input points. We call it "sensor fusion" or "data fusion" because it only really works when all of these components are able to share data with each other and agree on the accuracy of the conclusions.
The next step is to consider where the system must use AI to solve problems, such as: "What should I do with this information?" Do you want to turn left? Do you want to go straight? Do you want to turn right? What's the situation?" Analyze different transient factors, such as people, bicycles, or cars, and then make decisions and plans. Of course, the car takes some physical action, which in turn changes the environment, so the cycle starts all over again.
The real challenge, therefore, is a high level of interconnectivity, with the system's strength depending only on the speed and quality with which data is captured and processed. Then, when you add things like connecting to the cloud or other systems, you add external interconnects, which are also part of the interconnect solution. So it's a very complex distributed system with a lot of components, all in a very compact package. But what exactly is it bound together by? It needs to be built on a flexible, massively scalable IIoT framework to keep pace with competitors, industry standards, and many other variables.
ConnextDDS and
The concept of a hierarchical data bus
Supporting massive scaling is a core prerequisite for every highly autonomous system. This truth applies especially to the field of driverless cars, where a system that is truly ready to go to market is built in a limited test environment with enough complexity to make the best development teams lose sight of the other. In order to get to market and meet public demands in all media censorship and new testing scenarios, it is often necessary to add a whole new layer of mission-critical tasks to the system, which no one has been able to do so far.
When we get to that stage — getting a system to work reliably and into production — that's where we can help. Because RTI can provide a very reliable foundation on which customers can build their own software. We have been working in the field of autonomous systems for many years. Customers trying to do all the work themselves is likely to do more than half the effort, especially when RTI's expertise can be leveraged to meet difficult challenges such as software infrastructure and communications.
Our ConnextDDS software is a good example of this capability, as it uses a hierarchical data bus to manage communications. Hierarchical data bus is a concept and term proposed by the Industrial Internet Consortium (IIC), of which RTI is a member. We have written a number of documents and specifications for the IIC. One of the results of working with other companies was the creation of a hierarchical data bus that allowed for a differential treatment of control and information flows at different levels of the system. In addition to having global control, it allows "quality of service" to be set arbitrarily to determine how data flows between applications in different scenarios, including reliability, bandwidth, and latency.
This concept of a hierarchical data bus allows us to use the same standards throughout the ecosystem. We can customize different conditions and rules for data management for different parts of the system. This allows us to communicate between different systems in a very standardized way without having to add new protocols and gateways or other bridges. As part of ConnextDDS, the hierarchical data bus makes it easy to find these different data usage conditions, making data management reliable and repeatable.
Finally, what we want is to free up our customers' development teams and focus on developing the cars of the future
Source: AI "Automotive Manufacturing"
By Bob Leigh
Work unit: RTI Corporation
【Important Statement】This article is an original article and may not be reproduced without permission
Read more
【AI original】ZF future-oriented electric drive technology interpretation: 800V, SiC, electric disengagement device... Learn about a wave
Analysis of autonomous domain controllers and current developments
In-Vehicle LiDAR: Complete the complete range of ADAS and autonomous driving sensors
【Original】Autosar-based automotive electronic control system analysis and design
Please send an email to:
(Welcome industry insiders to contribute and share your articles with everyone)