The Internet of Autonomous Agents Part 1
A Quick Introduction to Cloud Robotics
Rapid developments in computational performance, wireless networks, and efficient data storage methods have given rise to the era of cloud computing, which enabled low-cost and easy access to cloud resources. The ubiquity of cloud use has now come to the point that not only humans and stationary things-as in IoT- but also robots benefit from these resources.
We define cloud robotics as an emerging technology in which robots utilize cloud resources such as computation, virtualization, parallelization, and storage to accomplish otherwise impossible tasks.
Offloading complex and computationally-heavy algorithms to the cloud have many advantages for any robot in terms of cost and efficiency. Currently, mobile robots and vehicles mostly depend on their onboard computational resources to execute tasks that leverage machine learning, probabilistic robotics, computer vision, and databases. However, we can only use a tiny portion of the applications of such technologies by built-in computing capabilities.
As listed in “A Comprehensive Survey of Recent Trends in Cloud Robotics Architectures and Applications,” an overview of the benefits of utilizing cloud robotics is as follows:
- Increased computational power and storage space.
- Solving computation-intensive tasks such as object recognition, pattern matching, computer vision, speech synthesis, and recognition much faster, cheaper, and in real-time by offloading them to the massively parallelized cloud infrastructure.
- Elastic and on-demand availability of cloud resources thanks to efficient virtualization and optimized networking.
- Applications such as Simultaneous Localization and Mapping(SLAM) and other perception systems that lead to the accumulation of a vast amount of data which cannot be stored or processed onboard.
- Having direct access to extensive data such as global and local maps, object and action data sets for various tasks such as navigation, detection, and grasping.
- Instant over-the-air updates ensure a fast and convenient way to sustain the modernization and security of the robot.
- Facilitation of cooperative learning between geographically distributed agents, and crowdsourcing intelligence robot to robot.
Recently, some of the largest cloud providers announced that they would be offering cloud services for robots. In the past six months, Google, Amazon, Microsoft, and Honda presented their cloud robotics solutions to the public with IBM and some other tech companies also integrating robotics to their services at some level. It is not a coincidence that these companies started offering cloud services for robots at around the same time. Three of these companies have already been offering some of the best cloud computing experience and leading the way for more than a decade. Providing a cloud infrastructure for robots is only a matter of ad hoc development for them.
Robot monitoring, teleoperation, simulation, and fleet management are the essential services that almost all of these tech giants will soon provide. Tasks like Simultaneous Localization and Mapping, online orthomosaic generation, 3D modeling, object detection and categorization, mission guidance and navigation are just some of the low-level applications that are going to be on the cloud for robotics.
Robotics-as-a-Service business model may lead to an exponential rise in the use of cloud-connected robots in many industries, but we are still very far from there. While these beloved tech giants undertake the massive mission of providing secure cloud infrastructure and front-end tools, we at Soar Robotics focus on how to deliver operational robotic systems to address industries’ pain points. For this relatively new paradigm to get adopted, what matters the most now is to create value out of these technologies by correctly allocating cloud and robot resources to accomplish tasks to solve real-life problems. Solving real-life problems requires a bulletproof architecture which benefits not only from the cloud but also the edge, making it possible to partition and distribute tasks across robots efficiently to create mission-driven, multi-agent autonomous systems.
At Soar Robotics, we aspire to a better future for autonomous robots. Our team has been working on solving some of the most time and labor consuming problems that many industries suffer. We have developed a fully automated, around the clock nested drone system from the ground up, for monitoring, inspection, and measurements in solar energy, agriculture, construction, and security operations. We are already leveraging the power of cloud robotics in our systems, lowering our hardware costs and downtime while increasing the quality of actionable intelligence that we provide for our customers. Utilizing our modular and inter-operable Intelligent Robotics Cloud, we are equipping the physical features of aerial robots with the cloud computing capabilities and cloud-powered artificial intelligence that result in smarter robots that cooperate to complete complex missions, collaboratively learn from each other and generate value for industries.
In the next article of this series, we will discuss ideas regarding different cloud robotics architectures, evaluate the role of edge computing in creating efficient autonomous systems, and dive deeper into our multi-agent autonomous architecture.
