Revolutionary Qualcomm RB6 AI Robot Cloud Accelerator

Revolutionary Qualcomm RB6 AI Robot Cloud Accelerator

Table of Contents

1. Introduction
2. The Qualcomm AI Accelerator
3. The RB6 Platform: A High-End Hardware Development Kit
4. The Power of Cloud AI-100
5. The QRB 5165: A Powerful System on a Chip
6. Vision Processing and Camera Support
7. Advanced Connectivity and Industrial Protocols
8. Energy Efficient AI Hardware Inspired by Neural Modulation of the Brain
9. The Stashing System: Reducing Energy Consumption in Neuromorphic AI Networks
10. Salience Labs: Marrying Photonics and Electronics in AI Processing
11. Conclusion

AI Accelerator from Qualcomm Packs Extra Punch for Autonomous Robots

Artificial intelligence (AI) is transforming the world of robotics, and Qualcomm is staying at the forefront of this revolution with its latest offering for autonomous robots and drones. The company has incorporated its Cloud AI 100 accelerator into a high-end hardware development kit called the Robotics RB6 platform, which serves as the "brains" of these intelligent machines.

Introduction

As AI becomes a driving force for robots, it is crucial to have powerful hardware to support their capabilities. Qualcomm recognizes this need and has upgraded its RB6 system with the Cloud AI 100, designed specifically for AI inferencing and data centers. This new hardware provides more than 10 times the performance of its predecessor, enabling smarter, safer, and more advanced robots.

The Qualcomm AI Accelerator

At the heart of the RB6 platform is Qualcomm's QRB 5165, a robotics-grade system on a chip manufactured on a seven-nanometer process. This powerful processor features the Cairo 585 CPU, the same one found in smartphones like the Snapdragon 865, boasting frequencies of up to 2.84 gigahertz. The CPU combines four high-performance cores for heavy computational workloads and four high-efficiency cores that help conserve a robot's battery life. The processor also contains Qualcomm's Hexagon DSP 698 with Hexagon vector extensions, as well as tensor and scalar acceleration engines. These components work together to deliver up to 15 trillion operations per Second for on-device AI and deep learning workloads, all while minimizing battery consumption.

The RB6 Platform: A High-End Hardware Development Kit

Qualcomm's RB6 platform is designed to be flexible and future-proof. It supports expansion cards that can be used to add support for upcoming releases and 5G networks. With 5G connectivity, robots powered by the RB6 platform can leverage ultra-fast millimeter Wave networks, enabling faster and more reliable communication.

The Power of Cloud AI-100

The RB6 system's AI engine is further enhanced by the Cloud AI 100 accelerator. Leveraging this accelerator, the RB6 platform delivers significantly more processing power for AI workloads compared to its predecessor, the RB5. With the Cloud AI 100, the RB6 system can achieve performance ranging from 70 to 200 tera operations per second. This immense processing power enables robots to handle complex AI tasks in real-time, making them smarter and more capable.

Vision Processing and Camera Support

To help robots Sense and react to obstacles in their environment, the RB6 platform packs in advanced vision processing capabilities. It can connect to up to 12 cameras via DPHY and 18 cameras via CPHY, with support for up to 7 cameras concurrently. This simultaneous processing capability allows robots to process up to 24 1080p video feeds at the same time, enabling real-time Perception of their surroundings.

Advanced Connectivity and Industrial Protocols

In addition to AI capabilities, the RB6 platform offers advanced connectivity options. Qualcomm's FastConnect 6800 system, with Wi-Fi 6 support, ensures fast and reliable wireless communication. The RB6 system is also designed to handle harsh industrial environments, operating in temperatures ranging from -30 degrees Celsius to 105 degrees Celsius. It can communicate over industrial protocols such as EtherCAT and Time-Sensitive Networking (TSN), making it suitable for a wide range of applications.

Energy Efficient AI Hardware Inspired by Neural Modulation of the Brain

Researchers in Korea have proposed a new system to reduce energy consumption in spiking neuromorphic AI networks. Inspired by the neural modulation of the brain, they have developed a technology that efficiently handles AI operations by imitating the continuous changes in the topology of the neural network Based on the situation. This approach, using neural networking techniques that mimic brain activity, can significantly reduce the high power consumption of AI accelerators in sensing and vision applications.

The Stashing System: Reducing Energy Consumption in Neuromorphic AI Networks

The research group from KAIST in Seoul has developed a learning method that directly implements neural coordination circuit configurations to reduce energy consumption. By using a self-rectifying synaptic memristor array and a stashing system algorithm, they were able to achieve a 37% reduction in energy without sacrificing accuracy. This stashing system is compatible with existing electronic devices and can be used in the design of next-generation semiconductor chips for AI.

Salience Labs: Marrying Photonics and Electronics in AI Processing

Oxford-based startup Salience Labs is pushing the boundaries of AI processing by combining the merits of photonics and electronics on a single chip. The team is currently prototyping a chip that harnesses the capabilities of light to execute operations. The photonic processor, capable of ultra-high throughput tensor processing, offers significant improvements over conventional electronic processor chips. It can process data at incredible speeds, enabling high-precision data analysis, and accelerating machine learning tasks.

Conclusion

Qualcomm's AI accelerator and the RB6 platform represent a significant leap forward in the field of autonomous robots and drones. With the integration of powerful AI hardware, advanced connectivity, and vision processing capabilities, these platforms empower developers and customers to build smarter, safer, and more advanced robots. The ongoing research efforts in energy-efficient AI hardware and photonics-based processing further contribute to the advancement of AI technologies, opening up new possibilities and applications in various industries.