.jpg)
Welcome everyone, and thank you for taking time to join this event. My name is Matt Weberge, solution architect at Advantech USA. Today we will discuss key factors in developing next generation AI robots and we'll share how Advantech solutions can accelerate their deployment by leveraging NVIDIA platforms. We've also invited an expert from Nvidia to explain how Jetsen and Orin modules help bring powerful AI solutions to the next wave of Edge AI and robotics. Industrial robots have become increasingly popular and are being used in a variety of industries, including manufacturing, defense, logistics, and healthcare. As companies have begun to understand the advantages robots offer. We've seen an increased investments in these technologies, as well as growth in different types of robots in the market segments they're used in. Future proof robots are expected to possess situational awareness, be capable of detecting and interacting with the environment in real time, finding the best matching route with pre programmed algorithms and have a certain level of self learning. There are four key requirements to meet these needs. The first is the ability for faster decision making to identify abnormal situations with task based or best route choices. Next, you'll need a real time operating system, leveraging data analysis and asset management. A reliable wireless connection is also necessary to facilitate smooth communication and remote management. Lastly, data processing utilizing AI at the edge is important to empower safe navigation. The power of AI technology in developing smart robotic platforms is advantageous, but something that may be added later along with other technologies as the platforms continue to evolve and improve. The demand of AI assisted computing is growing in the embedded market, especially with industrial robots. Most system integrators are including high performance GPUs and their robotics applications to enhance the AI capabilities. Most developers will face four challenges when integrating new GPU technologies into an industrial robot. These include having a diverse software development environment. It's a big investment to build the software team required to up your applications. There's also the unexpected maintenance costs that intermittent field failures will cause, and you'll also need to contend with a rapidly industry where time to market is a key factor in winning potential business. Finally, you need a scalable architecture that will offer you the flexibility to support new customer requirements. To help solve these challenges, Advantech provides comprehensive AI solutions based on NVIDIA Jetson and RTX products. Which are ideal for robotics applications. Our solutions include NVIDIA Jetson Edge AI systems and GPU accelerated platforms. On top of that, Advantech offers the ROS2 suite to accelerate ROS integration and our device on software to develop a large scale deployment and management capabilities. It's widely accepted that high end GPU cards are essential when processing sophisticated algorithms, but they aren't developed for the scarce conditions that may be exposed to when integrated into industrial robots. At Advantech, we've developed chassis solutions that protect these cards from shock vibration they'll be exposed to. The chassis and Advantech motherboards are also developed and tested to withstand high levels of ESD to help ensure uninterrupted we've also integrated efficient thermal solutions that help dissipate the heat generated from these powerful GPUs. Our forty years of experience industrial computers is crucial in protecting the NVIDIA GPUs that help drive today's robotic platforms. To support a variety of different AI computing performance needs. We offer a comprehensive product portfolio, supporting NVIDIA Quadro series, AI acceleration cards from the Compact MXM module up through full height dual slot GPU cards. These solutions have been NVIDIA qualified or certified to help ensure complete customer satisfaction, product reliability. We're excited to announce the newest members of our Edge AI solutions, which will be integrated with the NVIDIA Jetson Orin. Our EPC are seven thousand three hundred is a Jetson compatible barebone box PC which supports Orin NX or Orin Nano AI modules. With an extended operating temperature range and wide power input is an ideal system for industrial AGV and AMR needs. The new AIR-030 is a powerful and robust system supporting AGX Orin with AI performance up to two hundred and fifty trillion operations per second. This module comes standard with multiple ios such as Canvas, serial, and digital I/O, but can optionally be equipped to port power over Ethernet, 5G, WiFi, and ten gigabit Ethernet. It boasts more than eight and a half times of AI performance of previous generations and can fulfill a rate of diverse robotic application requirements. Now, I would like to pass the time over to our next speaker Chen Su, senior technical product marketing manager NVIDIA. He will elaborate more about the NVIDIA Jets and OR modules, and bring powerful AI solutions to the next generation of intelligent robots. Thank you, Matt. My name is Chen Su, senior technical product marketing manager, from NVDIA Jetson. Jetson is a software defined platform for AGI and robotics. There are three things you want to know about us. First, our hardware, edge AI and robotic system are driven by increasing number of sensors. This in turn is driving the need for higher performance compute in the system that integrate and process sensor data. Jetson platform, designed to be a small phone vector, high performance and power efficient that fuses data from many sensors to per receive, reason, and act in real time. Having a capable computing platform is necessary but not sufficient. The largest challenges by far develop and deploy an AI system resulted in the soft. Customers are looking to develop and deploy still embedded system need fundamental changes. You want to have reference software that became use or accelerated the life cycle of software management and development. Jetson provides full accelerated and application specific aid free more and models to make development easy and see full cost of ownership. Lastly, embedded and AI projects are complex and challenging. No one company is likely to be successful on their own. Justin offers a robust broad ecosystem to cover everything from sinisters or hardware to AI software in order to help our customers deliver a complete working system. Looking at the flu, Jetson family. On top, demonstrations six Jetson remodels, ranging from Internet AI to high performance. We call it autonomous machine. For the first time in Jetson in history, they make all products to be the same overhang, make it really easy for anyone to get started. Moreover, Jetson experiences a charm to keep in performance, benefiting from the ampere TPR connector. Put things into perspectives. The Jetson not only novel series set the new standard for Internet AI that offering up to forty tops at seven to fifteen horsepower budget. The Jetson Orin in the next series delivers up to a hundred tops at ten to twenty five watts per bike. The ATX owner's series offers a mind loading, two hundred seventy five tops server class performance at the edge, at fifteen to sixty horsepower budget. As we continue to innovate on the hardware, we keep the same software stack, JPEG five, for all the application migration and development, with upgraded OTA, security, and Bluetooth system. At GTC 2023, which successfully on shelf all JETSON known modules. Customers can go productions with more flexible options and are able to support, check out to Advantech, which include many great robotics full system powered by a mid adjustment layer. How to market is easier than. The next product in the JETSON family is our industrial SKU of Enziaxul, which provides higher operating temperature range, better shocking and vibration tolerance. We will launch the product in May as supported by JetPack 5.0 5.0.2 . As I mentioned earlier, Jetson is a software on assist platform, Velker, and a customer like us because of continued software investment. Let me break it down into three key building blocks. The first one is AI model development. We offer between AI model synthetic data generator and how to kits to reduce your time on AI model training by up to ten times. Moreover, Jetson have a Google bus application framework write it down different AI workloads. I see for robotics, is trained for intelligent video analytics, and revolve for conversational AI. Those not only provides you the easy to use library, but also improve your model performance by end to end hardware acceleration. I will give you a deep from the group or who box the annual in the next slides. Lastly, the JetPack SDK. The developer, it now has production ready, fully validated, and high quality menu system packaged up with up your favorite video AI compute that, like Buddha, Goodman, multimedia, visionary, to build the end to end accelerated AI pipeline. Let me bring everything together by showcasing you the power of combining the media, Jetson Orin, with our Advanced robotic SCA. IC Ross is a collection of hardware accelerated package. It's designed to make it easy for a raw lawyer to build high performance solution. The new acid ice cream recipes released, optimized while two nodes, asking for pipeline, which can be executed from Jetson Orin platform, and provides new DNN based gems for higher. On the table here, we listed many sample perceptually in IC Boston. A protect, dope, image segmentation, prosimity segmentation, still disparaging, the performance is measured under load, including message transport costs, in a hard CL, and for practical benchmarking, indicative of your real world performance. By using the latest ROS TP the Jetson AGXO on sixty 4G. We can achieve up to ten times for image segmentation. All of those made possible by eliminating offer, CPU overhead, and improving performance of hardware acceleration. Now let me pass down to Nathan and talk about Advantech innovation in robotics. Nathan will achieve. Thank you, Chen. You are insight and the expertise on robotics and HAI is variable. I believe in entire audience has much deeper understanding of the potential of NVIDIA Orin platforms and how they can be used in different applications. I believe we can all agree that the next-generation AI has significantly enhanced performance upgrades recently. It will impact a lot of applications and industrial solutions. Here, I'll take AMR as an example: AMR is usually several sensor types, such as lidar, ultrasonic sensors, and cameras for machine vision. This requires real-time image classification, object detection, and pose automation. The next Jetson Orin platform, deliver a large performance upgrade. Improving inferencing by 4 to 7 times, making AMR capable for more advanced tasks. The new gesture already performed, it will be a large performance upgrade, improving inferencing by four to seven times, making AMR capable more advanced tasks and more with greater flexibility. So if you are planning your next-generation robot, The NVIDIA Orin platform is definitely your ideal choice. Besides AI computing cores, there are many components in robot, all of which need to co-work harmoniously as a system in order to complete real tasks with precision and efficiency. For example, lidar distance sensors and 3D cameras are used as the eyes of robot, collecting and detecting objects in environment for navigation control. For cobot applications, the robotic arms need either a CANBus or PCIe interface to control arm movement. The number of components can easily be more than 15 for integration, and each component might have different interfaces. Which presents a great challenge for robot manufacturers. Advantech has many hardware platform solutions from board-level to complete systems with scalable AI computing performance, and plenty of I/O imterfaces with expansion flexibility. This makes it much easier for the developers do integrate peripherals. Our target is to help robot manufacturers develop a reliable and the advanced intelligent robots. With shorter development times by leveraging our hardware and software offerings. Developing a robot is a complex multi-level task. that requires expert knowledge in various domains, such as mechanical engineering, electrical engineering, software engineering, and many others. In addition to developing the robot itself, manufacturers must also make several important considerations for deploying and managing robot in the field. At the development stage, we provide a one-click installation software package to directly create the ROS & AI development environment. Through Advantech’s Linux BSP, it can help developers install and build an OS image quickly and easily. Besides that, we have also integrated Advantech DeviceOn features for Over-the-Air image and application provisioning and updates. For field site development, control nodes and edge nodes can empowered by DeviceOn for remote management and control. This make it easy to integrate with customer's fleet management system or other existing IT systems. With Advantech's solutions, we are not only providing you with hardware platform, but also application-oriented with software add-ons and services. Advantech’s AIR-030 and EPC-R7300 a double latest NVIDIA Orin-based technology, and their hardware design targets specific field side requirements, including wide range power inputs as well as they need for compact chassis and the industrial-grade operating temperature support via a fanless thermal solution. We are also considering robotic needs by adding application-focused I/O support for robotics peripherals, video capture, and wireless connectivity. In our software offering, which is based on NVIDIA, Jetson Linux, Advantech's AIM-Linux service comes with pre-integrate robotics peripheral drivers and optimized filesystem for developer. In the application development layer, we have integrated the NVIDIA ISSAC ROS SDK and the ROS2 framework. In the next two pages, I will go deeper to show how it's works and how it benefit developers at the software development stage. The AIM-Linux software service is dedicated to accelerating the software development of Arm-base platforms. It provides a flexible and modular framework aimed at the foundations of industrial markets, focused on long-term Linux BSP maintenance and oriented toward delivering longevity support for diverse Advantech Linux supported products. Advantech leverages AIM-Linux to deliver a verified solid foundation with unified embedded platforms, application-focused value-added add-ons, and SDKs that allow users to focus on vertical application development without expending resources and effort on platform integration. In BSP, Advantech’s long-term maintenance, and support policy offers regular BSP updates every half year. It fulfills fewer kernel updates and security considerations while design next-generation products or integrating new peripherals. As a primary platform for ROS, Ubuntu is support by JetPack for regular security updates empowered by NVIDIA. Regarding driver integration, Advantech provides peripheral support lists with verified drivers in BSP for robotic applications.These peripherals, including cameras, sensors and motor controllers are ROS-compatible for easy integration. Advantech also offers industrial-grade displays and WiFi and 5G wireless modules for fast system integrations. For more technical support needs, the AIM-Linux Developer Center is one-stop resource hub with comprehensive documentation and information for Advantech embedded computing solutions.This documents cover all releases, test commands, user guides, and the reference images to simplify your development process. Please scan the QR code on the slide to view our AIM-Linux Developer Center. If you have any technical questions, please feel free to post your question in our AIM-Linux community, where we are happy to share our experience and knowledge online. ROS, which stands for “Robot Operating System,” is an open-source software framework for robot software development. It provides a set of software libraries and tools for many tasks, such as communication, perception, and the control, allowing developers to focus on creating the unique features of their robots. ROS covers everything from drivers to state-of-the-art algorithms, and with powerful developer tools, which is what is essentially needed in robotics development, it is widely used in academia and industry for research and product development in the field of robotics. The Advantech ROS2 Suite offers a one-click installer and container-based software packages with pre-validated hardware platforms. These Suite also includes the SUSI API. SUSI is a suite of application interfaces that enables users to directly monitor and control digital I/O, I2C, watchdog timers and many other hardware interfaces. Additionally, industrial protocols and edge time-series database software are also pre-integrated. General ROS utilities like RVIZ, MoveIt, and Gazebo, which are also frequently used in robotics applications, are also pre-integrated into the ROS2 Suite. In addition, AI accelerator SDKs from silicon vendors like NVIDIA are also crucial, and so they have been consolidated into the ROS2 Suite to empower edge intelligence efficiency from native processing units inside of processors. This unified software design helps the developer to integrate ROS2-based applications and easily transfer it across various NVIDIA-based platforms. Advantech, they unlocks new possibilities in the field of AMR by integrating the ROS2 Suite into edge AI solutions. These power-efficient solutions comprise a wide spectrum of NVIDIA AI-native platforms in diverse products that facilitate flexible and easy application for end users seeking to develop and expand different AMR portfolios. The ROS2 suite is coordinated and proven software package built on Advantech’s AIM-Linux embedded software. It is designed to support ROS environments. Likewise, it delivers the highest degree of integration for positioning, navigation, and motion execution to empower users during robotic process automation. As a hardware-software integrated solution, Advantech’s edge AI products with the ROS2 Suite enable immediate development without wasted effort. In summary, Advantech’s robust, powerful, and unified solutions are vitally important to the AMR market. Next, we have short video clip recorded by our Advantech Japan team to demonstrate how to use Advantech hardware and software solutions to build up AMR autonomous navigation with collision avoidance. Let’s take a look. As embedded design expert, Advantech’s Robot Design-In Services will help you realize entire robot systems that are efficiently designed and can be built in no time at all. Apart from embedded hardware platforms, we also provide software, peripheral integration, and robot onboarding solutions. The complete design-in services cover operating systems, machine learning AI, the ROS2 Suite, and robot management tools, and all the software we’ve mentioned today are ready to engage with different types of robotics applications. If you are dealing with robot development right now or planning to do so in the future, our Robot Design-In Services can definitely help you during development. To conclude, I'd like to quickly recap Advantech’s value to the audience. First, Advantech provides a comprehensive product portfolio with the latest NVIDIA Jetson Orin and RTX technology for the next generation of intelligent robotics. Second, Advantech’s ROS2 Suite can accelerate your ROS-based application development. Finally, Advantech’s Robot Design-In Services can help you streamline your AI robotics development in terms of real-time OS, AI management software, and ROS integration specific to your project requirements. Advantech is ready to support your AI robotics projects. We currently have more than 95 sales offices across 28 countries around the world, so please feel free to get in touch with us for further discussion. Finally, thank you for joining us today. We look forward to hear from you soon.