Describe the role of feedback in a control system.


Teaching Note:

LINK Connecting computational thinking and program design.


Sample Question:


JSR Notes:

(Before starting with this one, note that it is one of these assessment statements which can be looked at in two ways, generally and particularly. So a question might be to generally describe the role of feedback in control systems, or it could be to describe the role of feedback in a particular situation, such as in an elevator. So be prepared for both approaches; on this page, the first part is more general, and the second part takes particular examples.)


General Description of Feedback

First of all recall the very beginning of 7.1.1; I'll paste it and its associated diagram again here:

control system is a device, or system of devices, that manages, directs or regulates the behavior of other devices or systems. This can be as simple as a remote controlled toy car, all the way up to extremely complex industrial control systems in manufacturing plants.

Often, (but not necessarily) an important part of the control system is feedback from the controlled device - these are called closed loop control systems, and utilize sensors for supplying the feedback; for example a rivet machine on a car assembly line that gives feedback about the resistance encountered, or the cruise control system of a car. Systems with no feedback from the controlled device are called open loop control systems; for example a simple hand held riveter, or the speed control of a remote control car.


wikimedia.org 680px-Ideal_feedback_model.svg.png

Simple diagram of a closed loop control system

In the above diagram, the feedback part is indicated by B (which would be the sensor) along with the arrows going from the output to the input.

"Feedback" is basically "communication", and it can often be seen as being two way. But in the case of a sensor control system we are talking primarily about what is diagrammed above. In a closed loop control system (i.e. one that uses feedback) a sensor continually monitors the result that is desired by the initial input.

So feedback occurs when the outputs from a control system are continually* monitored/sensed, and the data generated is used in some way, usually to keep the system in a stable state.


*Constant Feedback

Note that usually in sensor control systems, the data being monitored is either continually changing, or can change at unexpected times. So it is important that the control system constantly monitor the information on which it is basing its decisions.



It is most useful to think of specific control situations, in which real-time changes have to be continually monitored so that the resulting instructions can be maintained/changed/adapted based on the "live" situation "on the ground".


Example 1. Pressure Sensor of an NXT Robot (JSR point - at some stage change this to Java & FTC kit)

In the case of the pressure sensor attached to the NXT robot we looked at in class, it's transducer part is simply the little button which, when pressed in far enough, establishes a contact between two wires, and sends a current signal to the processor.

As part of the robot's control system, flowing current from this sensor is interpreted, when received, as a boolean 1. In the code run by the processor, while that signal from the sensor is 0 (i.e. current is not flowing), the robot moves forward at 100% speed, and when the signal turns to 1 (i.e. electricity is flowing) the robot goes backwards at 20% speed for one second - see below.

//RobotC code for NXT brick with a touch sensor attached to 
input 1, and two motors attached to inputs A and B. task main(){ while(SensorValue(touchSensor == 0){ motor[motorA] = 100; motor[motorB] = 100; } motor[motorA] = -20; motor[motorB] = -20; wait1MSec(1000); }
Movie of this Feedback to a Robot's Touch Sensor.


Example 2. Refrigerator

A refrigerator is able to keep a constant temperature to keep food safely cold through a feedback system of a temperature dial, thermometer sensor and associated microprocessor.

--> The initial temperature is set on the dial by the user
--> A thermometer sensor keeps track of the temperature of the fridge
--> Within a certain tolerance, when the sensor reports that the temperature has increased, this feedback results in the microprocessor turning the refrigeration unit turns on
--> And within a certain tolerance, when the sensor reports to the microprocessor that the temperature has decreased beyond the initial temperature as set, the refrigeration unit is turned off.

In this system the initial input of the user setting the temperature first affects the outputs (the microprocessor turns the refrigeration unit on). Then the outputs affect the inputs (refrigeration unit changes temperature reading of the sensor). We can thus say that the outputs affect the inputs; i.e., that this system involves feedback. The feedback keeps the system in a stable (in this case, constant, cold) state.


Example 3. The cruise control of a car

----------> EXCELLENT overall lesson on feedback and closed-loop systems <-----------
- using a car's cruise control as the main example
- a must-watch, even though it's 9 minutes

This is the system that allows the driver of a car to set a certain speed he/she wants to drive at, and no longer have to keep their foot on the accelerator pedal.

--> The initial speed is set by the driver when they reach the desired speed
--> A sensor keeps track of the speed of the car
--> Within a certain tolerance, when the speed decreases - for example because of climbing a hill - the accelerator is automatically physically depressed further, until the initial speed is once again resumed
--> And within a certain tolerance, when the speed increases beyond the desired speed - for example because of descending a hill - the accelerator is released bit by bit or the breaks are automatically applied until the initial speed is once again resumed

Example 4. ABS (Anti-lock Braking System) braking system of a car.

This control system will apply the breaks only until skidding takes place. Upon skidding (i.e. the brakes have locked because of the lack of friction between the tires and the slippery road) it releases the breaks. But as soon as the skidding stops, it applies the breaks again.

If you have never experienced it, in a car applying ABS breaks hard in a sharp stop, it feels like a pulsing of the breaks during a quick, intense deceleration. This on and off, on and off, repetition requires continual feedback to work properly, achieving a controlled stop in the shortest amount of time possible. And lest we forget to mention, saving thousands of lives every year.

The feedback is as follows:

--> The initial input of the brake (i.e. indication that the driver wants to stop or slow down) is simply the brake pedal being depressed by the driver
--> Sensed information about whether the brakes are locking or not is continually sent to the microprocessor. This is based on sensing a combination of non-turning wheels, and continued motion forward of the car
--> Upon receipt of locking event information, the microprocessor sends a signal to the brakes to be released completely or gradually gradually.
--> And upon receipt information that the locking has stopped, the brakes will be automatically applied again, either fully or gradually more and more
--> This constant monitoring and applying/releasing of brakes goes on until the vehicle has come to a complete stop, or the driver has released the brake pedal.

Note that feedback can occur in information-based systems as well. Often an output will have a result on further inputs. For example, the output of accepting an on-line booking for an air ticket will be to reduce the number of tickets available.


How to Handle Other Examples

For each situation, ask yourself two things:
1. What is being sensed continually, i.e. what information is being sent to the microprocessor?
2. What affect do certain changes in what is being sensed have in the control of the device/situation?

Describe the role of feedback in the following control system:

Automatic Doors: The role of feedback in an automatic doors control system is as long as the sensor does not sense an object within its range, the doors stay closed, and as soon as the feedback is received that there is something in the range, the doors open. There is also feedback from the inner part of the doors, that if there is pressure before they are fully closed (i.e. a person is stuck) that pressure feedback triggers the doors to re-open.

Heating Systems: The role of feedback in a heating system controlled by a digital sensor is that the sensor senses the temperature in the room, and when the temperature drops below a certain level, the heating system turns on, but when the temperature rises above a certain target level, then the heating turns off. The difference between the "turn on" and "turn off" temperatures must not be too close, otherwise the system would be constantly turning on and off.

Taxi Meters: In a taxi meter system which is managed by an odometer/timing system, the distance traveled is sensed/calculated in the normal way (by sensing the rotation of the wheels), and then that information is used to calculate the driving part of the fare (the fare may rise in certain situations, such as after midnight). Additionally, waiting time is measured and recorded and used to calculate additional charges for situations where the taxi driver has to wait for the passenger and/or they are stuck in traffic.

Elevators: Among the feedback systems of elevators, is constant monitoring of the amount of weight in the carriage; if it is over a certain per-determined safe weight, it will signal to the passengers in some way that the weight must be reduced before the elevator will move. Also, there is constant feedback from the proximity sensors in the door, which, when an object is sensed in-between them, the doors will open.

And so on... for the following, I'll just put in point form key sensors and control

Washing Machine... - weight controlling time of wash, - buttons pressed determining the kind of wash...

Domestic Robots... - pressure when hitting a wall etc., - the shape of the room for more complex "floor coverage" algorithms,...

GPS systems... Satellites: - exact location of each satellite itself, resulting in varying on-ground calculations. Ground GPS devices: - the signals from the three satellites being used, which allow the determination of position

Traffic Lights... - on-coming cars, via motion sensors, or weight sensors, control what color the lights should be

and other common devices...


Best content in compSci | Diigo - Groups