User:Whelanky/sandbox

The NASA/NBS Standard Reference Model (NASREM) is a Real-time Control System for robots originally made by NASA. NASREM is a complex action selection system which can sense, plan, and act within an environment. The system stands out as an example of world modeling method of selecting actions. NASREM uses hierarchical structures and node processing structures to process information. NASREM is can be an intelligent structure for including goals, memory, analysis, and implementation.

Overview
The NASA/NBS Standard Reference Model was first implemented in NASA's Flight Telerobotic Servicer The robot was designed to build and maintain parts of the International Space Station and required and intelligent and complex model that could function in a complex environment. In response, the NASREM architecture was created to process the information in a human like way. NASREM functions under the assumption that intelligence requires value based selection and model based intelligence by using a world model in its processing. This was meant to emulate the complexity of traditional thought process models. The NASREM model includes directed systems for sensing, planning, and acting, which were central concepts to human psychology and decision making. NASREM carries out these tasks with an organized system of similar, coded modules that are connected through coding structure. The overall shape of the structure is a rectangular matrix. The vertical exchanges of information represent information exchanges flowing into more decomposed tasks in regards to the goals. Horizontal exchanges represent information exchanges at a similar level of abstraction from goals. A human operator can interject at any module in the system, and databases are constantly accessed by modules.

Hierarchy
Modules in NASREM are organized in hierarchy that flows from abstract goals to physical implementation of the goals. This element is the part of the system that selects the actions based on their place in the structure. This is where the task decomposition elements break down the goal into smaller tasks, which are then lower on the hierarchy.

Task decomposition occurs in NASREM with repeating patters of job assignment, planners, and executor functions. The job assigners are the parts that break down the larger goals from previous memory. Planners select which of plans to be processed by executors from decomposed goals or memory. The executors initiate the next step based on the current world model and plans given by the planners.

Sensory processing in NASREM is an integrated with memory, world models, and input from sensors. Sensory information is predicted in comparison to the world model to try to model the world in real time. Sensors are variable depending on the robot.

The system's world model is an abstract construction of the environment around it and extra possibilities in it's memory. Most of this model comes from it's access to Global Memory. The rest is taken from sensory input from the environment, task execution, and planning networks. The world model works to abstractly predict, update, and see possibilities of what could happen in the environment.

Human Interface
A controversial component of the NASREM system is the requirement of a human interface. With a human component, a person can adjust the coding at any point in the hierarchy to smooth the computation process. The systems that operate under NASREM are not fully independent systems. However, this element allows for problems to be solved if a problem should emerge from the environment that the robot was not programmed to deal with.

Criticism
The NASREM architecture is often criticized for its complexity. The need to represent goals abstractly in real time requires massive computing power, memory, and technical programming. Engineers argue that a system does not need to have working memory or a representational model of its environment in  order to be intelligent. The NASREM architecture is however effective at dealing with complex situations that require dynamic solutions.