Jonathan asked me, if I would like to do a project with him on Braitenberg vehicles. After some research and reading the first couple of chapters in Vehicles: Experiments in Synthetic Psychology, I was hooked in. Here is the first version of a Braitenberg vehicle, powered with two RC-Servos and an Arduino as its brain.

Best of all, it needs no soldering, drilling or hot glue. And if you’ve played already with Arduinos, there is a good chance, that you have already most of the needed parts at home..

Braitenberg vehicles

Valentino Braitenberg developed a model of simple vehicles with sensors and actuators (motors) and interconnections between them. While the vehicles are extremely simple, the emerging behaviour is not. It is often interpreted as love, aggression or caution.

The easiest one is a light seeking vehicle. That’s like “hello world” in robotics. The sensors are affecting directly the motors. The right sensor affects the left motor and the left sensor affects the right motor. That means, if light shines on the right sensor, the left wheel turns. And if the light shines brighter on the right sensor, the left motor will turn faster than the left one and so the vehicle will turn towards the light source.

These kind of simple robots can be build with analog techniques alone, they don’t need a microcontroller. Think of two sensors feeding into two amplifiers that control the motors. The big advantage a controller brings in, is the possibility to rewire the connections between inputs and outputs in software. Even more complex functions for the interconnections can be reprogrammed easily.

Needed parts

• Arduino board
• small breadboard or prototyping Arduino shield
• 2 RC-servos, can be cheap
• 2 wheels, Solarbotics
• 2 light sensors, LDR (Light Dependent Resistor), e.g. CdS from Solarbotics, and 2 resistors
• 2 3-pin headers
• battery holder and 4 rechargeable batteries
• some rubber bands
• some wires
• paper clip

I am using a Boarduino here, that snaps nicely into the small breadboard. The two wheels are used for convenience. You could use any other type of wheels and attach them to the servos.

The two resistors have to match the LDRs to form a good voltage divider. Otherwise you get only a small range of values out of your sensors. Mine work great with a 10 k resistor.

The servos are hacked. Hacking servos means modifying them for continuous motion. A standard servo moves its tiny arm around from -90 to +90 degrees. But we want them to act as simple motors. There are quite a number of resources out there on how to hack a servo. One of the latest and very good documented one is from Tod. Check out his post about “Tiny Servos as Continuous Rotation Gearmotors”.

Tools

Nothing. Ha, no soldering iron, no drilling and no hot glue! Ok, you need a PC and an USB-cable.

Assembling

Attach the two servos to the breadboard by using 3-pin headers. Connect the red cable to VCC and the brown one to GND. The orange cable is used to send the control pulses to the servo motor. It is connected to Arduino pin 10 (left) and 9 (right).

The LDR and the resistor are forming a voltage divider. Connect the LDR to VCC and to a free socket on the breadboard. Now connect the resistor to GND and a free socket of the same row as the LDR. Next connect a wire from this row to the analog input pins of the Arduino (left to analog 0 and right to analog 1).

Now take the two servos, put them together and wrap a rubber band around them and the breadboard. Then attach the battery holder to the breadboard and fix it with another rubber band.

Use the paperclip or some other kind of wire to form a small hook. The hook should snap into place and holding the breadboard and the battery holder together. And it has a little notch that is used instead of a third wheel.

Now this tiny guy is complete.

Arduino Code

/*
 * Simple braitenberg vehicle
 * http://tinkerlog.com
 */

#include "Servo.h"

Servo leftServo;
Servo rightServo;
int leftValue = 0;
int rightValue = 0;

void setup() {
  leftServo.attach(10);
  rightServo.attach(9);
} 

void loop() {
  // sensor values between 50..900
  leftValue = (analogRead(0) - 50) / 50;
  rightValue = (analogRead(1) - 50) / 50;
  leftServo.write(89 + rightValue);
  rightServo.write(89 - leftValue);
  delay(10);
}

Yes, that’s all it needs. Only 25 lines of code. Including comments.

The analog values of the sensors are in a range between 50 and 900. So we take 50 as 0 and scale the value down.

The value you send to the servo is the degree it should turn to. From 0 to 180 degrees. At 90 degrees it is centered. For me, 89 is the value at which the left and the right servo stands still. If we add a value, the servo motors spins forward, if we subtract a value, it spins backward. The function for the right servo subtracts the values because it is attached on the opposite side.

You might have to write some simple sketches to evaluate the right values of the sensors and servo motors.

If you plug the USB cable into your PC, it may suck too much power because of the servo motors. You can power it with your external battery pack. Check your Arduino board, most have a jumper for external power supply. Or you unplug the servos for programming.

Play

Now switch him on and see if he finds some light sources. It works better, if the rest of the room is dark with only a single light source. I managed to tempt him with a flash light or a lighter. If he seems not to turn as much as he should, try to bend the light sensors more sideways.

This small bot is a bit shaky because of the rubber bands. But he is forgiving. And as he moves around underneath your table, you almost instantly think of him as something that has its own will. Even if you know, that it has only two sensors and two motors.

Links

• Wikipedia: Valentino Braitenberg
• Amazon: Vehicles: Experiments in Synthetic Psychology
• Video of Jonathan’s vehicles

As a demo, I built a “nervous” BlinkM. It is getting nervous, if something moves in its view. And the more it moves, the more it gets nervous.
It consists of my Boarduino clone, a Sharp GP2D12 range sensor and a BlinkM. The Arduino reads the range of the range sensor and computes a value for the nervousness.

Talk to the BlinkM

Interfacing an Arduino with the BlinkM is easy as the BlinkM comes with lots of sample code. You just have to copy an include file (BlinkM_funcs.h) and place it beside your Arduino sketch and include it. All the I2C stuff is taken care of.

After initializing the BlinkM you can send commands to it. You can, e.g.

• fade to a color, defined by RGB
• fade to a color, defined by HSB
• set fading speed
• write a script
• play a script

Code

My two sketches are really simple. One changes the color, the other changes the frequency. Here is the code for the first one.

/*
 * Nervous Blinkm
 * Change the color depending on range changes
 * 2008/03/14
 * http://tinkerlog.com
 */

#include "Wire.h"
#include "BlinkM_funcs.h"

#define RANGER_PIN 3              // pin for the sharp gp2d12 range sensor
#define DELTA_THRESHOLD 10        // threshold for the range sensor
#define MAX_RANGE_CHANGE 50       // max range change that can be detected
#define CALMED_HUE 172            // hue value for calmed state

int blinkm_addr = 0x10; // the address we're going to set the BlinkM to
int distance = 0;
int oldDistance = 0;
int nervous = 0;
int delta = 0;
byte i = 0;

void setup() {
  BlinkM_begin();
  BlinkM_setAddress(blinkm_addr);
  Serial.begin(19200);
  byte rc = BlinkM_checkAddress(blinkm_addr);
  if (rc == -1) {
    Serial.println("rnno response");
  }
  else if (rc == 1) {
    Serial.println("rnaddr mismatch");
  }
  BlinkM_stopScript(blinkm_addr);
  BlinkM_fadeToHSB(blinkm_addr, CALMED_HUE, 0xff, 0xff);
  BlinkM_setFadeSpeed(blinkm_addr, 0x10);
  nervous = 0;
}

void loop() {

  // do 4 samples of the distance to reduce the jitter of the range sensor
  distance = 0;
  for (i = 0; i < 4; i++) {
    distance += analogRead(RANGER_PIN);
    delay(10);
  }
  distance = distance >> 2;
  Serial.print("distance: ");  Serial.print(distance);

  // compute the distance change
  delta = min(abs(oldDistance - distance), MAX_RANGE_CHANGE);
  // sum up (only if delta is noteable)
  nervous += (delta > DELTA_THRESHOLD) ? (delta - DELTA_THRESHOLD) : 0;
  if (nervous > CALMED_HUE) {    // limit the nervousness
    nervous = CALMED_HUE;
  }
  Serial.print(", nervous: ");  Serial.println(nervous);

  // In HSB the hue 172 stands for blue and 0 stands for red.
  // So subtracting the nervous value of the initial hue (blue) results in:
  //   "more nervous" --> "more red"
  BlinkM_fadeToHSB(blinkm_addr, CALMED_HUE - nervous, 0xff, 0xff);
  oldDistance = distance;
  nervous = nervous * 0.9;       // reduce the nervousness with every cycle
  delay(50);
}

Demo

Unfortunately it is not possible to combine the two sketches into one to control the color and the frequency.

Conclusion

The BlinkM is a cute little circuit. It has some computing power on its own and a real bright RGB LED. It is easy to program and you don’t have to bother about PWM. You can even script it. On the other hand it could get pricy if you want to make something that needs more devices than a hand full.

Links

• BlinkM
• ThingM, the makers of BlinkM
• Shop a BlinkM at Sparkfun
• Programmable LED
• Sketch to change the frequency, nervous_blinkm_freq.pde

• 2 resistors 10k
• 1 resistor 100k
• transistor 2N3904
• 2 capacitors 0.1u
• electret microphone

Prototyping the amplifier

Putting things together on a breadboard.

Actually I had no 2N3904 around, so I replaced it with a BC337. The circuit is a emitter circuit with voltage degeneration (I dont know if that exists in english). I dropped the couple capacitor and took the signal right away at the collector.

Prototyping with the ATmega

The sound sensing is done with the ADC of the ATmega. A simple program reads the analog value of the amplifier over and over. If the value (loudness) exeeds a specific level, an LED is lit.

The schematic for rebuilding.

The code can be found here mic_sensor.c. It is just hacked together and has lots of room for improvements.

Video

Here is a video that I made. Has lousy quality, for both, video and audio.

It was easy and worked pretty well. I enjoyed looking at the LED responding to the music. I haven’t recorded anything with this amplifier, it might sound awfull. Next steps could be playing the sound back or be able to analyze the sound (FFT). And (re-)learning more on analog circuit design.

Links

• Simple preamp
• Audio Amplifiers