Sunday, March 2, 2014

Response to Control in an Information Rich World: The Puzzle of Feedback and Control

What exactly are feedback and control?

According to R. M. Murray, control is a way of modifying behavior in machines or other manufactured products based on feedback from the surrounding environment. A product achieves this feedback by first performing a task in the environment and analyzing the computational feedback it receives from sensors.

Control and feedback are a way of limiting uncertainties in a products performance. They allow a system to compare its current performance with a desired one so that adjustments can be made to appropriate variables in its performance to help it more closely resemble the ideal performance. Even is the environment in which the product is located is constantly changing and altering what the ideal performance constitutes, the control and feedback can help it achieve a closer approximation to ideal than the product could without them.

Simply put, control allows the system to correct itself when it is not performing as well as it could based on feedback on how it is currently performing.

Control and feedback are not just limited to electronics. Chemical and biological examples include the human brain and nervous system. There's a reason why a person quickly jumps and pulls their hand away from a hot stove.

"Control is information science" - R. M. Murray

Control can be traced back to the industrial revolution. If a train engineer had to make a certain deadline, and suddenly he finds himself burdened with a few extra cars, then he or she will have to change his speed in order to meet the deadline. The same would be true for a cart and mule, or a modern day tractor trailer. Fundamentally, control and feedback have remained the same, it's just the ways we received feedback, analyze it and act on have changed. Dramatically.

The one thing about this article that I did not understand was the concept of feedforward. It is defined as "actions taken based on plans", but to me that sounds like a machine without a control and feedback feedback mechanism. If a machine is running blindly, without any information about what's going on around it, then is it not technically acting 'according to plan'? That won't necessarily mean the plan will work.

I was VERY intrigued by the aerospace applications of feedback and control. From its early forms to modern day, feedback and control have become so sophisticated that they can be used to create unmanned aircraft, capable of targeting specific locations, if not people, and destroying them with little margin of error. The amount of sensors, decisions and calculations a single unmanned vehicle, like a drone, would have to use and make is baffling to me, and all in a series of a few minutes. Though the topic is controversial, the science behind it is breathtaking. Even though the article did not expressly talk about drones (given its age), it hinted at the possibility of such machines being within reach.


No comments:

Post a Comment