Robotic CKM Applications

/Robotic CKM Applications
Robotic CKM Applications2018-07-14T16:02:33+00:00

Reason not Rules

All models are designed from the standpoint of utility. CKM technology provides the flexibility to represent any concept that a human can comprehend – in software. Domain models can be tailored for any utility where humans employ their own knowledge.

Models supporting natural language comprehension need to be broad and embrace the concepts associated with the vocabulary humans use to converse within a specific domain. However, many applications require a deeper more technical model of a problem domain or a complex system but do not need natural language interfaces. For example, monitoring and control applications where the inputs are from instrumentation and the outputs are commands.

These applications embody human expertise in the control and operation of complex systems and devices, endowing them with an unprecedented degree of autonomy and intelligence. We call these applications Robotic CKM.

Other than the problems they are intended to solve, Robotic CKM applications bear no relation to the traditional rule-based expert systems of the past. Those systems reached conclusions by chaining individual rules together, so each assertion represented by a rule was locked into one or more chains of reasoning aimed at specific conclusions. The problem with that approach is the validity of a given rule or assertion varies with context. Rule-based systems are often termed heuristic systems for this reason, meaning rule-of-thumb. Heuristics may be useful in many contexts, but validity cannot be guaranteed. Traditional expert systems could only compensate by adding more and more rules. The result was systems that were expensive to develop and maintain, particularly in mission critical domains.

Robotic CKMs contain expert knowledge of the system and subsystems along with their static and dynamic properties as they are understood by human experts. As with all CKM applications, the model structures are contained within our epistemological kernel which defines multiple and dynamic contexts. The human expert describes the system or problem domain using New Sapience tools and techniques that do not require a mental paradigm shift from the familiar engineering domain to a rule-based or programming methodology.

The run-time software which reasons about the system or problem under consideration does so at a level of abstraction impossible to obtain with rule-based systems. The result a much higher level of automation and increased mission assurance at a much lower cost.
Incoming data can be a real-time stream, such as telemetry or static data files such as stored instrumentation readings. If the application is for control as well as monitoring, the system will send control signals back to the system.  Outputs to an operator, if present, will normally consist of high-level state or status information generated by reasoning about the device or system data.