Qualcomm Robotics RB5 Platform puts 5G, AI in developers’ hands

Qualcomm has been a pioneer in wireless telecommunications for 30-plus years. To maintain its spirit of innovation, the San Diego-based company now spends approximately $5 billion per year on R&D. It’s taking its deep-rooted expertise and applying it to other verticals, including robotics.

Under the stewardship of Dev Singh the last four years, Qualcomm has made major in-roads with the robotics development community. But today it took another major step in hopes of becoming the de facto development platform for robotics companies. Today it released the Qualcomm Robotics RB5 Platform, which it calls the “world’s first 5G and AI-enabled robotics platform.”

Qualcomm has been shipping processors in the robotics space for years, but the RB5 is its most advanced hardware platform yet. It is designed to help developers tap into the benefits of 5G connectivity and AI to create high-compute, power-efficient robots for a variety of industries and applications. Building off of the RB3 platform, which Qualcomm introduced in February 2019, the QRB5165 processor is at the heart of the new RB5. It integrates Octa Core Qualcomm Kryo 858 CPU, Qualcomm Adreno 650 GPU, multiple DSPs (compute, audio and sensor), and the new Qualcomm Hexagon Tensor Accelerator (HTA).

The RB5 platform can also support up to seven concurrent cameras for both indoor and outdoor settings. This will be useful for simultaneous localization and mapping (SLAM), object detection and classification, autonomous navigation and path planning.

On top of the heterogeneous computing architecture, the processor features the fifth-generation Qualcomm AI Engine that delivers 15 Tera Operations Per Second (TOPS) for running AI and deep learning workloads. For comparison, the RB3 delivered 3 TOPS, so the RB5 offers a significant jump in performance, especially for AI-powered robots. With support for 4G and 5G connectivity speeds via a companion module, the RB5 platform could pave the way for the proliferation of 5G in robotics.

Singh, senior director of business development and head of autonomous robotics, drones and intelligent machines at Qualcomm Technologies, told The Robot Report the RB3 will continue as a product. The RB3 is suitable for many robotics applications, while the “premium-tiered” RB5 is designed for those that need AI, machine learning and computer vision capabilities, rich and multiple cameras, extra processing power and 5G connectivity.

“We are providing the key building blocks for any type of robot, including anything from a $200 consumer robot to a $2 million industrial robot,” Singh said. “We have the best heterogenous compute platform. It’s the perfect brain for robots, and our platform is a Swiss army knife for robotics development.”

The RB5 platform is available in multiple options, with flexible chip-on-board designs and system-on-module designs. Singh said there is an option for extended lifecycle support up until 2029. “This helps companies future-proof themselves,” he said. “This is for critical devices that need to be in the field for a long time.”

RB5 Development Kit

Of course, the RB5 platform comes with a purpose-built robotics development kit. Singh said this allows robotics companies to “customize their autonomy.” It offers robotics developers flexible software capabilities, including support for AWS RoboMaker, Linux, Robot Operating System (ROS), and Ubuntu. The RB5 robotics development kit comes with pre-integrated drivers for various cameras, sensors and 5G connectivity.

For example, it includes out-of-the-box support for the Intel RealSense Depth Camera D435i and Panasonic TOF Camera to provide leading depth- sensing capabilities, but it can work with other depth cameras. All the processing from these cameras happens at the edge via the QRB5165 processor.

“Take the Skydio drone, for example. It has 7-8 cameras that take in data from all directions, and it processes that data at the edge using our processor,” said Singh. “The system needs time-stamped images from all directions to do mission-critical navigation. That’s how you accomplish autonomy.”

“We are excited about the promise that the Qualcomm Robotics RB5 platform holds for the  small Unmanned Air Vehicle (sUAV) industry,” said Abe Bachrach, chief technology officer and co-founder, Skydio. “As the leading U.S. manufacturer of autonomous drones, we leverage breakthrough computer vision and AI to create the world’s most intelligent flying machines. The cutting edge 4G/LTE, 5G, heterogeneous compute, AI, camera, and security technologies integrated into the Qualcomm Robotics RB5 platform will enable us to create the next generation autonomous drones.”

TDK adds a complete range of sensors for enhanced robotics applications for the RB5 Development Kit, including single and dual IMU, magnetometer, chirp sensor, angle sensors, pressure and temperature sensors. The development kit also provides support for OpenCL, OpenGLES, and OpenCV.

A development kit for the Qualcomm Robotics RB5 platform is available for pre-order now from Thundercomm.

RB5 platform could advance field robots & drones

Singh said the RB5 platform will help accelerate robotics adoption in a variety of segments. The graphic above breaks down the segments robots that are targets for the RB5, while the remaining consumer robots are opportunities for the RB3 platform.

The real promise of 5G is especially crucial for robots deployed outdoors. Low-latency is one of the main promises of 5G, while the low-power, high-compute at the edge is the benefit of the RB5 platform.

“A lot of folks push the cloud as a solution, but that doesn’t work all the time. There’s a time for edge and a time for cloud,” said Singh. “For data privacy and low-latency, on-board intelligence is paramount to making decisions. You don’t have the luxury of time with mission-critical systems to send info to the cloud, make a decision, and then send that decision back.”

Another opportunity that could prove fruitful for the RB5 is the defense industry. The U.S. Department of Defense (DoD) recently launched an effort that invests $11 million to identify and prototype new personal reconnaissance drones that give soldiers a situational advantage.

Partners look forward to Qualcomm RB5 platform

Qualcomm cited several studies in its announcement of the RB5 platform. One study projects robotics will be a $170 billion industry by 2027, while another said the 60 million robots deployed in 2020 will double in number by 2025.

Qualcomm said it expects commercial products based on the RB5 platform to be available in 2020. It already has 25-plus early adopters and 30-plus ecosystem partners for the RB5. The following endorsements of the RB5 platform could be a potential sign of where some of those products will come from.

“Today’s RB5 platform news continues our work with Qualcomm to help developers easily build intelligent robotics functions using AWS cloud services,” said Roger Barga, General Manager of Robotics and Automation Services, Amazon Web Services (AWS). “Our collaboration with Qualcomm is designed to give robotics developers a complete robotics development, testing and deployment solution, with the scale and flexibility of the cloud and powerful 5G and AI enabled hardware.”

“Open Robotics is the hub of a global community, creating and applying open software and hardware platforms for robotics. In our work we aim to lower the barrier to entry into the field for students, researchers, application developers, and entrepreneurs alike,” said Brian Gerkey, chief executive officer, Open Robotics. “So I’m excited to see the release of the Qualcomm Robotics RB5 platform, which will help to spur yet more innovation in robotics across a variety of application domains.”

“Today’s global events have brought the value of automation and robotics sharply into focus, so it’s exciting to see Qualcomm Technologies address such a broad array of robotics categories and applications with its enhanced platform, which powers the world’s largest fleet of autonomous mobile robots (AMRs) operating in indoor public spaces through our BrainOS software technology,” said Dr. Jean Baptiste Passot, vice president, platform and AI, San Diego-based Brain Corp. “Qualcomm Robotics RB5 platform will provide a big innovation boost to the industry at large, while enhancing San Diego’s reputation as a global hub for AI and robotics.”

“We believe the Qualcomm Robotics RB5 platform will offer a critical advantage for our next-generation unmanned systems,” says, Roger Wells, FLIR Vice President and General Manager of Unmanned Systems & Integrated Solutions.”

“The new Qualcomm Robotics RB5 platform is a game changer for drone and autonomous robotics, as its machine learning accelerators will allow adopters to replace manual inputs with AI,” said Lorenz Meier, chief executive officer, Auterion and chairman, Dronecode.

“ADLINK is very pleased to support the launch of Qualcomm Robotics RB5 platform. As a global leader in edge computing, ADLINK is committed to serving our customers across industries with cutting-edge technologies in AI, machine vision and Industrial IoT,” said Henry Hu, global business development director, ADLINK TECHNOLOGY INC. “We are excited to be one of the early adopters to embark on a series of new developments based on Qualcomm Robotics RB5 platform. By harnessing the power of the Qualcomm Robotics RB5 platform’s Qualcomm AI Engine and 5G connectivity, ADLINK will be able to significantly leverage our deep expertise in heterogeneous computing and industry-grade ruggedness, and ultimately enable a broad range of next-generation applications such as autonomous mobile robots, industrial robots, and unmanned aerial vehicles.”

Darin Andersen, chief executive officer and co-founder of NXT Robotics Corp., said the company leveraged Qualcomm’s flexible design options for the development and commercialization of its autonomous public safety vehicles. This allowed the company to rapidly move from new feature prototype to full commercialization. “The [Qualcomm Robotics RB5 platform] integrates key capabilities such as high-performance heterogeneous computing, 5G/LTE, the Qualcomm AI Engine for on-device machine learning and computer vision, hi-fidelity sensor processing for perception, odometry for localization, mapping, and navigation, strong security, and Wi-Fi connectivity,” he said.

“Siasun works closely with Qualcomm Technologies, [and] with the powerful computing capability of up to 15 TOPS based on the fifth-generation Qualcomm AI Engine and other advanced features, Qualcomm Robotics RB5 platform will give Siasun’s Pongbot ping-pong robot system more possibilities, including light-weight mobility, remote real-time interaction of multiple devices, real-time AI computing and feedback on cloud,” said Luo Yang, vice president, Siasun.

“The incredible amount of AI processing and 5G capabilities built into the Qualcomm Robotics RB5 platform will enable new autonomous drone experiences, such as navigating quickly through tight spaces while mapping the environment for objects of interest. We look forward to working with Qualcomm Technologies and integrating the Qualcomm Robotics RB5 platform to build off of our existing work with the Defense Innovation Unit to expand the capabilities of our sUAS platforms,”says ChadSweet, CEO of ModalAI.