RidgeRun is excited to share our team’s expertise on the Qualcomm® Robotics RB5™ and RB6™ Platforms in our recently published Developer Wiki. This comprehensive guide contains all the essential information you require to effectively utilize the boards, whether you are a developer or simply looking to experiment with this platform. It serves as the perfect starting point for your journey.
Nowadays, with the importance of beating time to market and the necessity of using powerful technology, many of our customers find themselves grappling with limited time and expertise to leverage the true capabilities of the hardware platform. At RidgeRun, we recognize the hurdles developers face and the importance of good documentation to overcome the challenges, that is why we made public our documentation on these Qualcomm® products.
To further support and enhance your product, RidgeRun’s diverse portfolio of solutions for audio, video and metadata which includes, GstSEI, In-Band Metadata, GstRtspSink and Embedded Overlay is now supported in the RB5 and RB6 platforms.
If you've just discovered these platforms and are seeking cutting-edge technology or if you are using them in your projects and need to find expertise to bring your innovations to life, you've come to the right place!
Unveiling the Qualcomm® Robotics RB5™ and Robotics RB6™
The Qualcomm® Robotics RB5™ and Robotics RB6™ Developer Kits are advanced platforms that empower the creation of cutting-edge robots, incorporating 5G connectivity, integrated artificial intelligence and machine learning functionalities, as well as state-of-the-art computing power and sensing technology. Let's explore some of their key elements:
AI Acceleration: The platforms are armed with a dedicated AI engine, enabling you to integrate complex deep neural networks.
GPU (Graphics Processing Unit) Acceleration: With robust GPU acceleration, the RB5 and RB6 enable high performance and efficiency in your graphics-intensive applications, critical for high-end computer vision and real-time processing tasks.
Computer Vision: The platforms offer FastCV, a computer vision library that contains numerous of the most widely used vision processing functions for real-time image processing and augmented reality capabilities. For example, face detection, tracking, and recognition, as well as gesture and text recognition.
GStreamer Integration: With GStreamer integrated into the RB5 and RB6, you can expand the possibilities of your multimedia applications, by leveraging GStreamer plugins for hardware accelerated video encoding and decoding, and running your custom machine learning models with hardware acceleration directly in your pipelines.
Showcasing Our Developer’s Wiki
We have created a wiki with easy to follow documentation on the characteristics and features these platforms have to offer, its hardware components, as well as, step by step guides ranging from how to flash the boards to creating applications and pipelines that leverage the several dedicated Hardware Units that enhance performance of applications. Here are some examples of the procedures you can find in the wiki:
Building an OS Image from Source: How to build an OpenEmbedded OS image with the Yocto Project and deploy it into the platforms, allowing you to align your operating system with your specific applications and requirements.
GPU Profiling: By implementing three GStreamer pipelines with OpenGL® plugins that utilize the RB5/RB6’s GPU we measured the percentage of GPU utilization.
FastCV Example Application: Our team has developed a FastCV example application, showcasing some of its computer vision algorithms and the ability of the library to accelerate the algorithm's processing time by choosing to execute them with the available hardware acceleration dedicated units within the platforms, such as the DSP (Digital Signal Processor) and GPU.
Video encoding and decoding: We have developed GStreamer pipelines for Hardware and software video encoding and decoding and measured CPU percentage usage.
AI Acceleration: We have developed GStreamer pipelines with TensorFlow models and measured their performance in terms of CPU cores used, frame rate and CPU percentage used, delegating the models processing to the DSP, GPU, and NPU (Neural Processing Unit). See an example of a running model in the video below.
Time-to-market is a critical factor in today's competitive landscape. We understand the urgency of bringing your robotic innovations to market rapidly. Our team of experts is dedicated to providing comprehensive support, helping you overcome technical hurdles, and ensuring that your development process is smooth and efficient. With our support and the Qualcomm® Robotics RB5™ and Robotics RB6™ platforms, your vision will be transformed into a reality. Let's embark on this exciting journey together!
What’s Next?
Find out more in our developer's wiki: Qualcomm Robotics RB5/RB6 - RidgeRun Developer
For technical questions or to ask for development support please send an email to support@ridgerun.com or send a message through https://www.ridgerun.com/contact.