Making Robots With The Arduino part 5.pdf

With The

Part 5 -Adding Sensors To Explore The World

A RDUINO

By Gordon McComb

You’ve spent hours designing and building your latest robot creation.

You bring it into your living room, fire it up, and step back. Behold!

Your beautiful new robot smashes into the fireplace and scatters itself

into tiny pieces over the living room rug. You remembered things like

motor speed controls, colorful blinky lights, even a synthetic voice, but

you forgot to provide your robot with the ability to look before it leaps.

your ArdBot (or other Arduino-based automaton) the

sense of touch and light. Augmenting these basic

senses are methods to detect objects in proximity to

your robot — seeing what’s there without having to actually

bump against them.

Proximity detection forms the basis of collision

avoidance — how to keep your bot from crashing into

things in the first place. Collision avoidance takes many

forms. Some of these techniques are designed to detect

objects very close to the robot, while others are made to

detect objects several feet away.

In this installment, you’ll learn about two popular forms

of proximity detection — ultrasound and infrared — and

how these low cost sensors are interfaced and used with

the Arduino.

The State of the ArdBot

differentially-steered robot based on the popular Arduino

Uno and compatible microcontrollers. Here’s what past

installments covered:

• Part 1 (Nov ‘10) introduced the ArdBot project, along

This article series has been on the construction and use

of the ArdBot (see Figure 1 ) — an inexpensive two-wheeled

FIGURE 1. The ArdBot Arduino-based expandable robot,

shown with this month’s enhancements: a rotating turret,

ultrasonic distance ranger, and infrared proximity detector.

SERVO 03.2011 43

Making Robots

I n last month’s installment, you learned how to give

www.servomagazine.com/index.php?/magazine/article/march2011_McComb

Where Have All the Sharp

Distance Sensors Gone?

switches with polling and hardware

interrupts to detect physical contact

with objects, and ways to use

photoresistors to steer the robot by

light.

some models of their infrared distance

sensors, allowing the stock that remains

“in the channel” (still held in reseller

inventory) to be depleted. New versions

are in production, but they may not be

universally available.

You may have noticed that several of

the Sharp infrared distance sensors have

become hard to get from some sources —

most particularly from Digi-Key, who at

one time was a major seller of the things.

These sensors are discontinued and/or

not available as new stock in retail

channels.

However, most of the sensors can

still be purchased from a number of

specialty online shops, such as SparkFun

and Pololu, both of whom carry large

inventories of several Sharp models.

What remains of the current stock may

not be useful for creating a new mass-

market product, but there’s plenty still

around for us robo-tinkerers.

Sharp has introduced a relatively

new style of sensor — the GP2Y0D805

and GP2Y0D810 — that are smaller and

less expensive. The ‘05 has a set 5 cm

proximity range; the ‘10 has a set 10 cm

range. Output is a digital LOW or HIGH.

This new class of sensor comes in a

DIP-size package, but it requires some

external components. Online retailers

such as Pololu (see the Sources box)

offer the sensors on a breakout board for

easy use in your projects. Even with the

addition of the breakout board, these

sensors are roughly half the cost of their

predecessors. They also have a much

improved response time: over 350 Hz

(350 updates per second), as opposed to

about 25 Hz of the older sensors.

Be sure to check out the

previous four issues of SERVO

Magazine for more details. This

installment describes three

important robotic functions:

programming an ultrasonic

transducer to accurately measure

the distance to objects; how to use

a Sharp infrared distance module

for monitoring proximity; and ways

to add a rotating turret so the

ArdBot can scan the room to look

for things nearby.

In the next issue, we’ll conclude

with putting all the pieces together,

combining what you’ve learned to

create an autonomous room

wanderer that’s able to seek things out, investigate its

environment, and follow the direction of its master —

that’s you.

with the popular Arduino microcontroller and basic

programming fundamentals of this powerful controller.

• Part 2 (Dec ‘10) detailed the construction of the ArdBot,

using common materials including PVC plastic and aircraft

grade plywood.

• Part 3 (Jan ‘11) covered the Arduino in more detail. It

also examined the ins and outs of programming R/C

servo motors that provide the locomotion for the ArdBot.

• In Part 4 (Feb ‘11), we learned about using bumper

About Non-Contact

Near-Object Detection

Last time, we learned how leaf switches are used as a

form of touch sensor to detect contact with objects.

Contact detection provides an immediate signal that

something looms directly in the way.

Non-contact detection senses objects without

having to hit them first. Near-object detection does

just what its name implies: It senses objects that are

close by, from perhaps just a breath away to as

much as eight or 10 feet. These are objects that a

robot can consider to be in its immediate space;

objects it may have to deal with, and soon. These

objects may be people, animals, furniture, or other

robots.

By detecting them, your robot can take

appropriate action which is defined by the

programming you give it. Your bot may be

programmed to come up to people and ask for their

name. Or, it might be programmed to run away

whenever it sees movement. In either case, it won’t

be able to accomplish either behavior unless it can

detect the objects in its neighborhood.

There are two common methods of achieving

near-object detection: proximity and distance.

Sources

If you’d like to build the

ArdBot, be sure to start with

the November ‘10 issue of

SERVO Magazine for Part 1 of

this series. Also check out the

following sources for parts:

Acroname

www.acroname.com

AdaFruit Industries

www.adafruit.com

HVW Tech

www.hvwtech.com

Prefabricated ArdBot body

pieces with all construction

hardware; 360 degree rotation

sensor turret:

Jameco

www.jameco.com

Budget Robotics

www.budgetrobotics.com

Lynxmotion

www.lynxmotion.com

Arduino Resources:

Mark III Robot Store

www.junun.org

Arduino

www.arduino.cc

Pololu Robotics & Electronics

www.pololu.com

Fritzing

www.fritzing.org

Robotshop

www.robotshop.com

Online Retailers of Arduino,

Sharp sensors, and/or

ultrasonic sensors:

SparkFun

www.sparkfun.com

• Proximity sensors care only that some object is

within a zone of relevance. That

44 SERVO 03.2011

F rom time to time, Sharp discontinues

is, if an object is near enough to be

considered important. Objects beyond the

proximal range of a sensor are effectively

ignored because they are not seen. Out of

view, out of mind.

• Distance measurement sensors determine

the space between the sensor and

whatever object is within detection range.

Distance measurement techniques vary;

almost all have notable minimum and

maximum ranges. Few yield accurate data

if an object is smack-dab next to the

robot, or very far away.

FIGURE 2. The Sharp infrared distance

modules rely on the displacement of

reflected light across a linear plane to

detect variations in distance.

Collectively, these sensor types are often

referred to as rangefinders, though only a

device that actually measures and reports

the distance of the covered range is a true

rangefinder.

Among the most common proximity and

distance measurement detectors used in

robotics are ultrasonic transducers, and

specialty infrared sensors made by Sharp.

Depending on the design of the specific

sensor, either can be used for proximity or

distance measurement. In practice however,

the Sharp IR sensors are best suited for

proximity, and ultrasound sensors are the

ideal choice for measuring distance. That’s

how these two detectors are used in the

ArdBot.

FIGURE 3. The analog output of the

Sharp GP2D120 sensor is a non-linear

voltage of between (approximately)

0.25 volts and 2.55 volts.

Using the Sharp

GP2D120 Infrared

Detector

The Sharp GP2D120 is among a series

of remarkable distance measuring devices

originally intended for industrial control.

They’re common finds in amateur robot

projects. These sensors rely on the

displacement of reflected light across a linear

sensor (see Figure 2 ).

Here’s how they work: A beam of

modulated infrared light from the sensor

illuminates an object. The beam reflects off

the object and bounces back into the sensor.

The reflected beam is focused onto what’s known as a

position sensitive device, or PSD. The PSD has a surface

whose resistance changes depending on where light strikes

it. As the distance between sensor and object changes, so

does the linear position of the light falling on the PSD.

Circuitry in the sensor monitors the resistance of the PSD

element, and calculates the distance based on this

resistance.

Among the Sharp detectors currently available through

FIGURE 4. Connection schematic for the Sharp GP2D120,

showing a transistor used to turn the sensor on and off.

retail channels are two general types:

• Distance judgment sensors provide a simple digital

(LOW/HIGH) signal that represents whether an object is

within detection range. That range is set at the factory,

and depends on the specific model of the sensor.

Common distances are five, 10, 24, and 80 centimeters.

SERVO 03.2011 45

FIGURE 5. Breadboard view of

the Sharp GP2D120-to-Arduino

connection. Note the pinout and

wiring order of the GP2D120 sensor.

Important! The wiring diagram relies

on existing connections on the

breadboard. See Part 2 of Making

Robots with the Arduino (Dec ‘10)

for details.

closer than a few inches and up to

several feet.

For example, with the GP2Y0D810 sensor, the output is a

digital LOW when an object is within its 10 cm proximity

range, and HIGH otherwise.

• Distance measurement sensors provide an analog voltage

that’s more or less proportional to the distance from the

sensor to the detected object. The voltage output is non-

linear, as shown in Figure 3 . These detectors work over a

span of minimum and maximum distance, usually no

Distance judgment sensors

are ideal for interfacing with

simple electronics, as they don’t

require analog-to-digital

conversion. Since the Arduino Uno

is equipped with a six-input ADC,

we can use either type. For the

ArdBot project, I’ve selected a

GP2D120 which has a range of 4

cm to 30 cm (about 1.5 inches to

12 inches).

The distance is reported as a

varying voltage, from approximately 0.25 volts (no

detection) to 2.55 volts (detection at minimum distance).

That’s according to the spec sheet, but know that there can

be a normal variation of a few tenths of a volt from one

sensor to another.

While the GP2D120 is capable of reporting distance

with acceptable accuracy, for the ArdBot I’ve elected to use

it as a “multi-zone” proximity detector. That is, instead of

hassling with converting its analog voltage to

some quasi-precise distance measurement, for

the ArdBot the GP2D120 will instead simply

indicate when an object is within preset zones.

The ArdBot relies on a separate ultrasonic

rangefinder for accurate distance measuring

and as a secondary near-object detection

check. More about the ultrasound rangefinder

in a bit.

See Figure 4 for a schematic diagram for

connecting the GP2D120 to the Arduino.

Figure 5 has the same circuit, but in

breadboard view. See that I’m being clever

here, and I’ve added a small 2N2222 NPN

type transistor in order to turn the GP2D120

on and off.

The transistor is an optional enhancement,

but there are a couple of reasons for it. First,

like all of the Sharp sensors, the GP2D120

takes constant measurements — about 25 a

second — as long as the device is powered.

Current consumption can go as high as 50

milliamps when no object is detected. While

that’s not a huge current demand, it’s

unnecessary power consumption when the

sensor is not actually being used.

Listing 1 - SharpIR.pde.

const int irCtrl = 12; // Digital pin D12

const int irSense = A5; // Analog pin A5

int distance = 0;

void setup() {

pinMode(irCtrl, OUTPUT);

digitalWrite(irCtrl, LOW);

Serial.begin(9600);

// Use Serial Monitor window

}

void loop() {

Serial.println(irRead(), DEC);

}

int irRead() {

int averaging = 0;

// turn on, wait 250 ms to completely stabilize

digitalWrite(irCtrl, HIGH);

delay(250);

// Get a sampling of 5 readings from sensor

for (int i=0; i<5; i++) {

distance = analogRead(irSense);

averaging = averaging + distance;

delay(55);

// Wait 55 ms between each read

}

distance = averaging / 5; // Average out readings

digitalWrite(irCtrl, LOW); // Turn sensor off

return(distance);

}

46 SERVO 03.2011

Second, each time the sensor takes a

new reading there’s extra line noise

induced into your robot’s power supply. By

turning the sensor off when it’s not

needed, the noise is completely removed.

(You can also help filter the noise by

adding some decoupling capacitors across

the +V and ground power connections, as

close to the sensor as possible. Try a 47 µF

electrolytic capacitor and a .1 µF ceramic

capacitor. Be sure to observe correct

polarity of the electrolytic capacitor.)

Listing 1 shows a sketch that

demonstrates how to use the GP2D120

with the Arduino. This is an example sketch

only; in the next installment, you’ll see how

the GP2D120 can be implemented for

object seeking and avoidance, as the

ArdBot is set loose in your living room and

left to discover what’s around it.

A couple of things to note in this

sketch:

FIGURE 6. Connection schematic

for the Devantech SRF05 ultrasonic

ranging module.

• To make a reading, the sensor is turned

on and then allowed to settle for 250

milliseconds (ms) before taking a

reading. The datasheet for the GP2D120

indicates a much shorter delay upon

startup, but I’ve found the longer period

is often necessary to avoid spurious

reads.

• The sketch takes five “samples,” each

separated by a delay of 55 ms. The five

samples are averaged to remove possible

incorrect readings due to momentary

glitches.

• After the sensor is read, it’s turned back off again.

FIGURE 7. Breadboard view of the

Devantech SRF05-to-Arduino connection.

the wiring order is “corrected” to place the +V power lead

in the center. In this arrangement, damage is less likely to

occur if the connection is flipped. I have shown such a

cable in the breadboard wiring view in Figure 5 . Regardless

of whether you use a cable type that re-arranges the wiring

order, be absolutely sure to observe correct polarity . Or

poof goes your sensor.

Open the serial monitor window to observe the actual

values reported by the sensor, converted from an analog

voltage to 10-bit digital values (0 to 1023) by the Arduino’s

ADC. Note that because the sensor does not output a full

five volts when an object is closest, you won’t get the full

1,024 steps. Minimum values are about 40 to 60; maximum

values are in the 625 to 675 range, depending on the

sensor and the reflectivity of the object (dark colors tend to

produce slightly lower values). Anything outside these

ranges can indicate a spurious reading.

Important! Take note of the connection diagram for

the GP2D120. Sharp uses a polarized JST connector where

the +V power lead is on the outside and ground is in the

middle. This is potentially dangerous when the other end of

the wiring is non-polarized, as is often the case with

connectors on general-purpose microcontrollers.

Numerous sources (such as Lynxmotion) sell adapter

cables that go from the JST locking connector to a standard

three-pin 0.100” female header. On many of these cables,

Using the Devantech

SRF05 Ultrasonic Ranger

Ultrasonic distance measurement —also called ultrasonic

ranging — is now an old science. Polaroid used it for years

as an automatic focusing aid on their instant cameras. To

measure distance, a short burst of ultrasonic sound (usually

at a frequency of around 40 kHz) is sent out through a

transducer; in this case, the transducer is a specially built

ultrasonic speaker. The sound bounces off an object and

the echo is received by the same or another transducer (this

one is an ultrasonic microphone). A circuit then computes

the time it took between the transmit pulse and the echo,

SERVO 03.2011 47

Plik z chomika:

Inne pliki z tego folderu:

Inne foldery tego chomika: