SoylentNews

Anonymous Cow Herd writes:

If you fancy yourself as a systems administrator, perhaps you have had the opportunity to administer multiple machines, simultaneously. Perhaps you've had experience with interactions between multiple operating systems, or with struggles to keep multiple machines synchronized.

Cluster administration is a special discipline. When administering a cluster, it is best to keep Sun Microsystem's motto, "The network is the computer", foremost in mind — a cluster IS a computer; it is a computer that is running a task that involves such complex and demanding processing tasks that the different processes that form this 'machine' run on separate-but-connected computers. Clusters are often distinguished by some sort of interconnection — networked file systems, shared memory and machine-spanning inter-process communications.

Throughout this work one encounters issues related to parallelization, as some tasks can be done in parallel, and some cannot. The entire purpose of a cluster is to enable an otherwise sequential task to be worked on, where possible, in parallel, and it frequently involves additional programming and architectural design to allow this to happen in as close to real time as is possible.

This brings us to robots. A robot is a cluster! When I grasped that, I fell in love with robots and I started looking for jobs involving robotics. The processing requirements involved in movement exceed the computing capacity of a single computer, and for this reason, it is common to have one computer dedicated to moving each appendage, dedicated computers for complex tasks like parsing visual input or controlling 'hands', and often yet another computer acting as the master kinesthetic controller over the entire 'infrastructure', in something approaching real time.

So, now, we introduce a cluster of robots. Above, I mentioned parallelization. The issue becomes tangibly visible when one has a cluster of robots cooperating in order to achieve some sort of a task. All of the requirements for interconnection and intercommunication that manifested themselves previously in the architecture of parallel processing and clustered architectures become explicitly visible when one watches, say, a cluster of octocopters juggling balls!

Which brings me to this article: 'Coordinating complex behaviors between hundreds of robots', https://www.sciencedaily.com/releases/2020/07/200701125453.htm :

In a new paper published online on April 29 in the International Journal of Robotics Research, Zavlanos and his recent PhD graduate student, Yiannis Kantaros, who is now a postdoctoral researcher at the University of Pennsylvania, propose a new approach to this challenge called STyLuS*, for large-Scale optimal Temporal Logic Synthesis, that can solve problems massively larger than what current algorithms can handle, with hundreds of robots, tens of thousands of rooms and highly complex tasks, in a small fraction of the time.

In closing, I want to point out that if the United States really wants wants to Make America Great Again, they can start by hiring those thousands of older engineers with experience supporting multiprocessing clusters, and train them in the cloud-based technologies that have superseded them, instead of leaving them to sit at home and watch TV.

~childo

Journal Reference:
Yiannis Kantaros, Michael M Zavlanos. STyLuS*: A Temporal Logic Optimal Control Synthesis Algorithm for Large-Scale Multi-Robot Systems:, The International Journal of Robotics Research (DOI: 10.1177/0278364920913922)

Original Submission