Intelligent Autonomy: For Complex Systems Only?
Updated: Jun 15, 2019
Autonomy is the ability of a system to achieve goals while operating independently of external control. In a natural way, we associate this concept with self-driving cars, unmanned military rovers or hi-tech robotic assistants. And we are correct. However, before going on reading this article, keep in mind that the technology described here has side applications in other industries as well, like pharmaceutics, mining, oil & gas, and more. We will come back to this later.
So an autonomous system operates with no control from the outside. How can this be accomplished, in terms of structure, components, and generic pieces? What does an autonomous system look like? Well, first of all, such a system requires a degree of intelligence. This means it must be able to represent and understand the past and the current state of the world, as well as reasoning about it. And that’s only the beginning because autonomy implies not only knowing about the environment but also self-awareness. So if we are to deem a system as autonomous, then it has to be self-directed and self-sufficient.
In collaboration with NASA* we have been building an autonomous system, generically depicted in Figure 1. The diagram shows an Intelligent Autonomous System Architecture (IASA). As the reader can see, there is a central Thought System. Here is where the thinking begins. The Thought System is nurtured by the surrounding, nearly independent subsystems, but it also yields thoughts for them. And more important: contemplation, as well as self-awareness/self-consciousness, takes place here, too. This thinking core is needed so that our autonomous fellow can:
learn from the environment and from itself
design desirable scenarios to pursue
plan on how to achieve its goals
execute such plans
Figure 2 shows the big blocks of NPAS (NASA Platform for Autonomous Systems), yet in more detail than in Figure 1. One of the inputs is the specific mission to be accomplished, usually a set of systems to keep healthy. These pieces of information, along with preprocessed data from sensors and components, are fed into the IASA. Inside it, the pieces are brought together to make NPAS aware of the situation, the goals and the available strategies for action, so that it can then build a suitable plan to achieve the goals, which is itemized at a pertinent level, and then it is scheduled and executed. Needless to be said, the system runs in real time and adapts to the circumstances as they change.
Is there something like this for the rest of the industries?
We might think that only the very techie industries (as space exploration and self-driven vehicles) are candidates for using autonomous operations, but the truth is that subsets of a whole IASA can be used in most industries either to keep mission-critical systems healthy or to optimize production under changing, complex operating conditions and without the human-error factor. For all industries need to evaluate, learn and plan, and then to execute the plan against a dynamic, ever-changing environment. And all industries benefit from self-awareness and proactivity of a decision maker system that can always explain the reason for its choices.
We are in the knowledge age, branded as the autonomous knowledge systems age. The experience gained at building intelligent autonomous systems for hi-tech companies can be and must be brought to the commercial arena so that all industries can handle the overwhelming complexity of a world which increasingly presses for optimization, unmanning, and smarter treatment of processes as the price tag for surviving.
D2K Technologies (www.d2ktech.com) is a tech minority-woman owned company, founded in 2014, with its main office located in Oceanside, California, working in close proximity to other high technology companies and several front-line research organizations.
*References to NASA in this article do not imply in any way an endorsement by the agency.
Information about the NASA Platform for Autonomous Systems can be found here: https://techport.nasa.gov/view/94884