Handing over human tasks to robots has been on the horizon for quite a long time. Such robots are named service robots. They are either helping us in our personal tasks such as vacuuming or cleaning chores, mowing our lawn; or in our professional tasks such logistic robots for moving cargos, delivery in hotels, guiding us in shops; inspection and maintenance in hazardous areas of nuclear reactors, space missions, shipyards [2], sewers [3], and mines; search and rescue missions [4]; heavy construction [6] or agricultural projects [5]; as well as conducting surveillance and home security.

A combined application of next generation of sensors, communication technologies, radar, laser-scanners, cameras, positioning systems (GPC), geographical information systems (GIS), and functionalities of detecting and avoiding obstacles, recognizing faces, navigating using a map, has made all these state-of-the-art robotic systems possible. However, these specially manufactured appliances must operate in our complex environment that is not designed for them, and they transmit, process, and control real-time information to accomplish their fully automated tasks flawlessly. This results in the need of advanced sensors, processors, and actuators, that all end up in very expensive robots.

In this post, the obstacles facing the robotics industry is provided in brief, and it is discussed how cloud robotics can be used to overcome these issues.

Robots have new capabilities without having new hardware. Since new technologies in Artificial Intelligence (AI) and computer vision is emerging and is being applied in robotics everyday, a one-time purchased robot hardware might be totally deficient in a very short time. Robots are not only about mechanical hands anymore. They perform many intelligent tasks nowadays and their processing and storage requirements would need to change with their new applications. This is one of the shortcomings of robotics industry as one will get stuck with the hardware-depended robot capabilities they have purchased. Either the customers should continue using their robot without acquiring the new features, or they shall enter a marathon of buying and upgrading the features with the new applications arising. Nonetheless, Cloud Robotics enables improving the functionality of simple robots by relying on the clouds computing as an extra source of memory and processing. Hence, employing the physical features of the usual robots in addition to the great virtual innovations of cloud computing results in smarter robots, with great tolerance to possible intelligent adjustments.

Extensive computation is off-loaded somewhere else to be done faster with lower power consumption. The AI and Computer Vison are undeniable elements of robotics application these days. There has been lots of interests in AI applications such as Manufacturing Automation Protocols (MAP), Simultaneously Localization and Mapping (SLAM), and Object Recognition (OR) [8]. These applications require more agile and complex processors than usual robots to implement their real-time and heavy computational functionalities. As such work-loads exceeds the capacity of usual robots, supporting the AI functions demand for equipping and upgrading the current robots with high-priced processors and their prerequisite essentials. Moreover, applications like grasping and mobile navigating would need access to huge databases of 3D CAD models and navigation maps, respectively. Providing such storage on-board, if feasible, would impose great cost and frequent need in updating the database. So, the direct and indirect cost of implementing on-board computation and storage is a challenge in the robotics industry [9]. Moreover, by offloading the heavy computation from the robot to a server, the robot power consumption decreases by far, and the computation of the robot can be used for its other capabilities.

C2RO Object Recognition product demo, all the computation is on the cloud server, and only uses a very light single-board computer (i.e., Raspberry Pi 3).

Centralized processing gives accumulative insight on sensor data. With the growth of robotics applications, there comes the massive number of sensors and the flow of huge data acquired. The large data availability is useful in machine learning purposes; however, there is a need for an upper-hand system to refine the raw data received from all the sources. Hence, a centralized server can function as both the storage and processor; however, such a server requires a constant investment and upgrading occasionally, to fit the systems’ new applications. With the flooding of data such as video, images, or maps, to or from robots, it is needed to have a service team always present to monitor the databases and processing centralized servers to avoid overwhelming them. Cloud robotics can be used as the centralized processing and storage technology behind the usual robots. The customers would have their robots connected to the cloud as the center of their data, or code to strengthen their operation with intelligent sensor analysis insight on the accumulated data.

Multiple robots collaboration either for efficiency or complimentary views. The individual robots in isolation cannot basically have a global view of the complex environment to operate complicated tasks. This would require expensive and consistent updating of their features. So, the extensive requirement for the current generation of robotics is to communicate to the other robots, share knowledge, learn collaboratively, and even share skills. In other words, multiple robots can overcome individual inaccuracies due to observation noise or lack of global view while learning cooperatively. The multi-robot applications would require more executive computation, communication, collaboration, and centralized administration. These are not possible unless huge cost-intensive renovations are implemented in the current infrastructures. To enable such collaboration, Cloud robotics facilitates reliable and agile connection and sharing of robot resources instead of relying on standalone robots. Therefore, robots will follow their tasks while cooperating through clouds to expand their processing features.

In a nutshell. To use existing non-necessarily-intelligent robots with limited memory and computation in high-tech applications, there is a need for utilizing a more comprehensive framework. Cloud robotics as in offloading the intensive computations from robots to clouds provides such infrastructure virtually on clouds. The robotic industries can use whatever processing and storage they need through cloud robotics without worrying about changing or upgrading their local servers and sub-systems. Robots can communicate and share their data collectively on clouds through their wireless connections to the cloud. Moreover, they can implement their real-time applications relying on the parallel computational resources on the clouds. Moreover, sharing the data, access and updating of the remote libraries of images, maps, and object data would be possible. Hence, cloud robotics can overcome the current issues in the robotics industry and pave the way for more penetration of robots in our daily lives while increasing the quality of it.


[1] Evan Ackerman, “Fetch Robotics Introduces Fetch and Freight: Your Warehouse Is Now Automated,” in IEEE Spectrum, April 2015

[2] D. Alonso et al., “Experiences Developing Safe and Fault-Tolerant Tele-Operated Service Robots. A Case Study in Shipyards,” in Service Robot Applications, Aug. 2008, pp. 159-182

[3] A. Ahrary, “Sewer Robotics,” in Service Robot Applications, Aug. 2008, pp. 287-308

[4] H. Zhang, “A Novel Modular Mobile Robot Prototype for Urban Search and Rescue,” in Service Robot Applications, Aug. 2008, pp. 215-233

[5] S. M. Pedersen, S, Fountas, S. Blackmore, “Agricultural Robots — Applications and Economic Perspectives,” in Service Robot Applications, Aug. 2008, pp. 369-382

[6] Bock Thomas, “Service Robotics in Construction,” in Service Robot Applications, Aug. 2008, pp. 383-400

[7] J. Ryu, B. Yoo, T. Nishimura, “Service Robot Operated by CDMA Networks for Security Guard at Home,” in Service Robot Applications, Aug. 2008, pp. 383-400

[8] B. Kehoe, S. Patil, P. Abbeel, K. Goldberg, “A Survey of Research on Cloud Robotics and Automation,” in IEEE Trans. On Automation and Eng., VOL. 12, NO. 2, April 2015

[9] G. Mohanarajah et al., “Cloud-Based Collaborative 3D Mapping in Real-Time With Low-Cost Robots,” in IEEE Trans. On Automation and Eng., VOL. 12, NO. 2, April 2015

Submit a Comment

14 + 5 =