The M3-Neony and M3-Synchy were developed as baby bots aimed at testing machine learning software, and specifically to take a look at fine motor skill development. The hardware on this adorable little bot are some typical cameras, a microphone, gyro, accelerometer, and tactile sensors.

I heard about the M3–neony and M3-synchy through this Engadget article but I was disappointed the coverage was so scant. When I began blogging for GoRobotics, I mentioned briefly my loved for HCI, and in particular human-robot interaction – naturally, this article inspired me enough for a second article today. But, as I was excited reading about it, it looks like the article only mentions briefly the research goals of the bots. There is, however, a lot of information about what was used to make them for you gearheads out there. I’m going to comb to find the Japanese lab site if I can, in the meantime here is what’s available so far:

This article at Plastic Pals seems to have more detailed specs on these two robots. The article is long, but features more detailed specs on the bot:

[...] it is 50cm (19.6″) tall, weighs about 3.5kg (7.7 lbs) – about the size of a newborn.  A pair of CMOS cameras for sight and microphones for hearing, as well as gyro and accelerometer sensors, and tactile sensors provide various feedback. The robot has a total of 22 degrees of freedom, powered by high torque (41kg/cm) servo motors sold by Osaka-based robotics company Vstone.

The main focus is on facial expressions and arm gestures, so it is an upper body robot only, with 17 DOF (2 eyes x3, neck x3, waist x2, 2 arms x3), measuring 30cm (12″) tall and weighing 2.5kg (5.5 lbs). The head is equipped with a single wide-angle lens CCD camera, two microphones, a speaker, and 15 LEDs which cause the robot to blush bright red.  Combined with object recognition, speech recognition, and speech synthesis, the robot will be able to communicate in a variety of ways.  The chest and arms appear to be based on Vstone’s Robovie-X hobby robot kit.

If anyone finds out more about what kind of tactile sensors are involved, I’d love to hear about it. Tactile sensors aren’t something I hear about a lot and I’d love to put together an article on what’s out there.

You can catch a video here, and do check out the Plastic Pals article – they have a great gallery of these baby bots.

Gåågle – It’s not as weird to pronounce as you’d think. It’s actually pronounced like Google and you’ll begin to see why soon enough. Gåågle Bot is a modified remote-control Roomba that bears a webcam, fueled by real-time AJAX calls that zips around taking pictures and indexing the real world as it sees it. Vacuum, index. I love efficiency!

!

The name Gåågle Bot is a play on the words gå and google bot. The Swedish word for go is gå. Googlebot, is the name of Google’s web indexer. If you don’t know what Google is, you are either lying or out of luck. Hence Gåågle Bot is a “going” indexer, indexing the real world around us while vacuuming your home at the same time! Can’t find that library book that is due tomorrow? Relax, just gåågle it!

Excited about this bot? Head over here and give it a try. There is also a pretty nifty video as well showing the bot in action. The main site has all of the components listed, the source code, and other tidbits to get you started building your own remote-control crawler.

The meat of today’s article is MIT’s MeBot.

MIT's MeBot

MIT has a pretty established humanoid robotics lab, meaning they’re at the forefront of our latent dreams to one day have cyborgs and robots walk the streets with our fellow man. (Call it whimsy, call it crazy, but I’m looking forward to an increasing number of robots in society. ) Anybody interested in robotics already knows of the legacy that MIT has for it’s robotics development, including Kismet – a rather impressive early attempt at robot-human social interaction (you can find more about Kismet here), and Cog – another human-robot interaction experiment that followed the reasoning that Cog should be able to learn from interacting with humans (more information about Cog here). MeBot comes to us from the Personal Robotics Lab.

Telerobotics is the area of robotics development concerned with – you probably guessed it – remote-control robots. The overarching idea of the field is that work needs to be done at a distance in some situations in life, and telerobotics is here to aim to answer those challenges.

The robot was presented at the Human-Robot Interaction conference in Osaka, Japan. Putting an OQO atop for a head plus some gesturing arms into the mix, it adds depth to the notion that you could really be there, and with a decent range of motion, rolling down the halls of MIT. Remotely. Via a robot.

The proposal here is that this mode allows the user to be more engaged through the movement of the head and arms. The head tracks  the face of of the user so that it can ‘look around’. The arms are moved by a set of hand-operated controls.

Gulper AUV Sub-Aquatic Robot

The group explains that it has ‘trained’ the robot to retrieve the highest-quality information back to them.

“We tell it, ‘here’s the range of tasks that we want you to perform’, and it goes off and assesses what is happening in the ocean, making decisions about how much of the range it will cover to get back the data we want.” says Dr Maughan of MBARI.

The Gulper AUV is used to help scientists keep tabs on various algae. In particular, these scientists are keeping watch for algae blooms that could means problems for the ecosystem.

It used to be the case that a ship would be sent out for a whole day every few weeks to retrieve the kind of information that the Gulper AUV can nab in one of its trips. They just take it out to the harbor, and away it goes on its mission. Around twenty-four hours later, it comes back, they hoist it away, and analyze the results.

The biggest flag to go off in my mind is that this must require some interesting exploration and path planning algorithms to deal with an undersea environment. Taking a look at MBARI’s website, the Gulper AUV is equipped with four sonar that operate simultaneously to provide a fantastic map of the sea floor in high resolution.

The multibeam sonar produces high-resolution bathymetry (analogous to topography on land), the sidescan sonars produce imagery based on the intensity of the sound energy’s reflections, and the subbottom profiler penetrates sediments on the seafloor, allowing the detection of layers within the sediments, faults, and depth to the basement rock. All components are rated to 6000 m depth. The vehicle is launched on programmed missions and runs on its own battery power until it returns to the ship, as programmed, for recovery – MBARI AUV Mapping Page

Head over to the article at BBC to hear an audio snippet about the Gulper AUV. it’s about halfway down the page. If you think that’s cool, then you’d also better head over to the AUV’s home page at MBARI to check out the technical goods.

In what has been an ongoing controversial move in the United Kingdom, police forces all over the nation will be able to draw on unmanned air drone robots for surveillance support. The units are remote-controlled and equipped with thermal imaging units, and they’ll set you back about $30,500. So far there is only one unit seeing action in the UK, and it’s already getting publicity for helping the police do their job.

The Merseyside police who happened to be lucky enough to have one of these $30,500 drones flicked on the thermal imaging on a tip that a suspected car thief was somewhere in the neighborhood. They managed to pinpoint the suspect from about three hundred meters away, and their actions also eventually led to the arrest of a second suspect shortly thereafter. Sky News has the coverage over here.

A young man was caught and arrested for breaking a law, which makes this a good day for robotics, and a good case for robots in a pragmatic, practical role. Still, speculation considers the increased use of robots within the police and military to be walking a rather fine line for safety, especially if future units are armed and are expected to operate with any sense of autonomy. Wired has an interesting article detailing the possible ways that police drones could be armed in the future.

A couple interesting submarine launched UAVs, one by Raytheon and another, VOLANS, built by a German company, are featured in this Register article. The Submarine Over the Horizon Organic Capabilities, or SOTHOC, built by Raytheon, is launched out of the waste disposal lock of a submarine. SOTHOC then decents to a preset depth where it rises to the surface and launches a unmanned flying vehicle to gather data. The UAV can relay the data back to the sub via antenna, or if the sub whishes to remain anonymous the data can be relayed via satellite back to the US. This system allows a submarine to lauch an UAV while remaining submerged, in contrast to the VOLANS, which launches via a mast attached to the robot. The VOLANS functions as a mobile periscope for the sub.

• soldering iron  ( here’s a helpful tutorial on soldering )
• electronic solder
• diagonal cutter
• Mini glue gun
  

• 2x – small 1.5 Volts motors
• 2x – small paperclips
• 2x – big paperclips
• 2x – batteries AAA or AA
• 1x – battery holder AAA or AA
• 1x – 2 cm of heat shrink
• 1x – wooden pearl  (for the caster)
• 1x – meter of electric wire
• 2x – Sub-mini lever SPDT switches
  
  

Most of the components can be bought for much cheaper at Digi-Key, Jameco, or similar. At Solarbotics you can find the dual AA battery holder and the Mabuchi motor. You can find these components at any good electronic store.

]]> http://www.gorobotics.net/articles/robots/how-to-build-a-simple-robot-beetle-robot/feed/ 117 http://www.gorobotics.net/articles/robots/how-to-build-a-simple-robot/ http://www.gorobotics.net/articles/robots/how-to-build-a-simple-robot/#comments Sun, 29 May 2005 19:30:39 +0000 admin http://www.gorobotics.net/wordpress/?p=431 This article lists some ideas for starting a simple robot. A good read for beginners.


This was orginally conceived as a multipart series from Mr. Craig
Gardner, but he took our money and ran. This is all we got. Lesson
learned.

If you like this article, please check out this. How to Build A Simple Robot, The Beetle Robot.

This is the first of a multi-part series on how to build a simple
robot. I will not be covering a lot of theory but instead will cover
the nuts and bolts of building a robot. In this first installment I
will cover some of the different options in building this robot.

Platform:

You have several options for the base material.

1. Wood Depending on the size of your robot plywood may be a good
choice it is inexpensive or free and easy to work with simple tools.
You can get 1/4" and 3/8" plywood at Hardware, Home Improvement, Craft,
or Hobby stores. A good place to look for free material is at a Cabinet
shop they have lots of scrap too small for their use but perfect for a
small robot.
2. Aluminum Light weight and moderately easy to work can also be found
at Hardware, Home Improvement, Craft, or Hobby stores. You should
always be very cautious when working with aluminum as edges can be very
sharp and should be sanded or filed to round the edges.
3. Plastics Acrylic or Plexiglas are both easy to work and can also be
found at Hardware, Home Improvement, Craft, or Hobby stores. High-speed
tools should not be used, as they will melt the plastic. When cutting
or drilling use low speeds.
4. Old CD’s These can be easy to find most people get them in the mail
from AOL or Earthlink instead of throwing them away you can use them to
make a robot. They can be a little on the brittle side so go easy when
you do any cutting or drilling.

I will be using Plexiglas from my local home center but you can use whatever material you want.

Motors:

1.Servos You can use servos for easy to get gear
motors. Hobby shops will usually carry several sizes and brands. You
will need to modify them for use there are many sites on the net with
different methods; the one I like is detailed at the PARTS website at http://www.portlandrobotics.org/.
Servos already have all of the control circuitry built in and are easy
to control they have 3 wires signal, +, and ground by pulsing the
signal line you can go forward, stop, or backward. Servos are probably
the easiest and cheapest way to go but may not be the best for you.
2.Gear motors These are available from surplus stores or hobby shops
some people modify servos and remove the electronics to use them as
gear motors. Gear motors will require control circuitry normally an
H-bridge to allow forward and reverse motion and in some cases braking.
Gear motors can give the greatest flexibility but at a higher cost
compared to servos. Another source for gear motors is the toy section
of your locale department store. What you want are the radio control
toys that have differential steering, meaning they have a separate
motor for each side. To turn left they go forward on the right motor
and turn off the left motor and do the opposite to turn right. The
really good cars will turn right by going backward on the right motor
and forward on the left this will allow a vehicle to almost turn on a
dime. If you use gear motors you will need to devise a method to mount
the wheels I will be using a toy for my example.

Wheels:
You can use wheels from toys or you can buy wheels from a hobby shop
they have pneumatic or foam wheels for model aircraft that are very
easy to use.

Power System:

Batteries You will need to decide what type of batteries to
use. It can quickly become very expensive replacing batteries.
Rechargeable batteries are best; there are a number of different types
to choose from. Electronic supply stores or Hobby shops are good places
to look, you will need batteries and a charger to charge them with.

Power supply We will need a voltage regulator to drop the
voltage from the battery to the 5 volts needed by the Microcontroller
and other parts of the Brain for the robot.

I will cover using two different voltage regulators both are
available FREE from National Semiconductor as samples. Each part has
its good and bad points.

LM2825 Integrated Power Supply 1A DC-DC Converter is a
complete switching power supply on a 24 pin DIP although a little large
it requires no other components and has an efficiency of 80%. It does
require at least 7 volts on the input but your batteries will last much
longer than with a linear regulator.

LM2940 1A Low Dropout Regulator is a linear regulator in a TO
220 package it requires a couple of filter capacitors it is not as
efficient as the LM2825 the big up point is it only requires 5.5 volts
input to give a regulated 5 volts out.

Tools and equipment required:

– Soldering iron (about 18W with a fine bit) and solder
- Safety glasses (wear them at all times)
- Long nose pliers
- Side cutters
- Wire strippers (optional)
- Hot glue gun and glue sticks (or an epoxy)
- 3 large capacitors (e.g 4700 µf)
- 1 efficient motor (e.g. portable CD player, Walkman, etc. motor)
- FLED Solar Engine

1. Attach the three capacitors to the motor body using a hot glue gun (you can use epoxy-resin but it’s easier with hot glue). Apply the glue to the lighter colored stripe on the capacitor body (this stripe identifies the negative lead). Align the capacitors vertically so that the bottom of each capacitor is approximately 4 mm below the bottom face of the motor.

2. Bend the three negative leads together and solder them as shown in the photograph.

3. Construct a FLED solarengine.

4. With the two transistors resting on the tops of the front two capacitors solder the emitter leg of the 3904 to a negative capacitor lead and the emitter leg of the 3906 to a positive capacitor lead (the 3906 is a PNP transistor).

5. Now solder the black (negative) motor lead to the protruding collector leg of the 3904 transistor (to which the resistor is attached) and the red (positive) motor lead to any one of the positive capacitor leads.

Congratulations you’re almost there !

6. From the two brass coated paper clips supplied construct a ring with a diameter of approximately 85mm - 90mm. Bend the clips with a pair of long nose pliers (wear safety glasses when you’re doing this) and also trim to length with the pliers. Solder together the ends of the wire to form the ring.

7. Attach the ring to your Trimet using the positive leads of the three capacitors. Form a hook with the leads to secure the ring prior to soldering. You will probably need to extend the lead of the front right capacitor (see photograph) – use a piece of the brass coated paper clip that you trimmed off in the previous operation.

We’ve got the traction, the fuel tanks, the engine and the sensor now all we need to add is the fuel converter and the drive wheel.

8. Solder two short (30mm long) pieces of the PVC coated multi-strand wire to the solarcell. Solder the red lead to the positive pad and the black lead to the negative. Take care when doing this – do not over heat the tinned pads on the solarcell. These are easy to solder because there is already a generous amount of solder on the pads. Solder the red lead so that it runs parallel to the ‘Panasonic’ text on the cell and the black lead so that it runs perpendicular to the ‘Panasonic’ text as shown in the photograph.

9. Solder the leads attached to the solarcell to the capacitor leads – red lead to a positive capacitor lead (any of those attached to the sensor ring) and black lead to a negative capacitor lead (the lead nearest the light stripe on the capacitor). Secure the solarcell to the body using a little hot glue (or a little epoxy-resin). Do not allow the solarcell pads to touch any of the component leads.

10. Cut a short section of hot glue stick (about 6mm – 7mm long) with a pair of scissors or a craft knife (take care!), melt a small hole in one end of the piece of glue stick (not in your finger – it hurts!) and push the ‘wheel’ onto the motor shaft. Position the ‘wheel’ so that there is between 5mm – 7mm from the bottom of the ‘wheel’ to the bottom of the capacitors. This will allow the Trimet to lean over onto two of the capacitors at the appropriate angle. You can adjust this dimension later to produce the optimum performance. Ideally you want the Trimet to skid along in a straight line – have you seen the James Bond film where he drives a car on two wheels down a narrow alley - like that!

Oh and by the way you’ve finished.

Place your Trimet (wheel down) on a smooth surface (plastic table top, piece of polished timber, etc) in sunlight (doesn’t have to be full sun) and wait for her / him to trigger (every 3 – 5 seconds in full sun). In the absence of daylight you can also use a halogen lamp or an ordinary bulb (although it will take longer to trigger).

Editors Note: TotalRobots is a British based robotics company. You can visit their website and order some of their products off their website. The above tutorial is Copyright © Total Robots.

Spencer Scott has written another excellent article about building a BEAM Symet. You can read his article and get some other ideas.

1. Take the 3904 transistor and using a pair of long nose pliers carefully bend the collector leg (that’s the one on the far right) so that it is perpendicular to the transistor body as shown in the photograph. Then again using long nose pliers carefully bend the emitter leg (the one on the far left) first perpendicular to the base leg (the middle one) but in the same plane and then back up towards the base leg as shown in the photograph. If you’re not sure about this take a look at step 3 and 4 it might be clearer when you can see the assembled engine.

2. Take the 3906 transistor and bend the emitter leg (the one on the far right in the photograph) first perpendicular to the base leg but in the same plane and then over the top of the transistor body. Then take the base leg and bend it perpendicular to the transistor body as shown in the photo.

3. Hold one of the transistors with a pair of long nose pliers so that the base leg of the 3904 is next to the collector leg of the 3906. The legs should overlap by about 6mm - 8mm. With your other hand pick up your soldering iron (which should be hot !), clean the tip and then melt a little pool of solder on the tip. Apply the tip of your iron to the two legs of the transistors to be joined for a second (no more !). The legs will be heated and the solder should flow between them easily. Do not overheat (it may damage the transistors). Continue to hold the transistors together for about two seconds until the solder has solidified.

Now wipe the sweat from your brow!

4. Solder the 2.2k resistor between the collector leg of the 3904 and the base leg of the 3906 as shown in the photo. This is best done by first ‘tacking’ both ends of the resistor to the transistor legs with a small pool of solder on the iron and then applying additional solder to make a more secure joint, once the initial ‘tack’ has solidified. Using a pair of side cutters trim the excess lead from the resistor.

Almost there !

5. Carefully bend the negative leg (the shortest one and also the one nearest the ‘flat’ on the FLED body) of the FLED at right angles to the positive leg, as shown in the photo.

6. Slide a piece of heat shrink tubing over the FLED and using a lighted match shrink the tubing over the FLED body. Whilst the tubing is still hot pinch the end opposite the leads to seal the tube, then cut off any excess tubing.

7. Solder the negative lead of the FLED to the emitter of the 3904 (the one that is bent backwards) and the positive lead to the base of the 3906 (to which the resistor is also connected)

8. Trim off the excess leads : FLED negative, FLED positive and 3906 base – as shown in the photo.

Well done you’ve finished !

Editors Note: TotalRobots is a British based robotics company. You can visit their website and order some of their products off their website. The above tutorial is Copyright © Total Robots.

The symet for some reason always tends to be a three fold rotation symmetrical robot, but I like to push the envelop a little. The Quaret, hence its name, is a four fold rotation symmetrical robot. It’s very small, very active, and a lot of fun to watch. So enough chit chat lets find some parts!

List of parts
(most are available at DigiKey)
-1 x cassette tape player motor (or equivalent)
-1 x Solar panel (i use the Panasonic sunceram 37×33mm)
-Coil of heavy gauge wire (copper is good, but anything will work!)
-1 x 2.2k resistor
-1 x CMOS 1381J (The letter denotes the voltage it is triggered at, try some higher or lower voltage ones if you want!)
-1 x 2N3904 NPN transistor
-1 x 2N3906 PNP transistor
-4 x 1000uF capacitors (these can be substituted for any rating wanted, just remember, the more capacitance the longer charge)
-small diameter heat shrink (for motor actuator)
-4 x small washers (you’ll see)
-plenty of electrical tape!

Now go find those parts!!

Now that you have the parts

So you finally got all of the parts, good! It’s a lot easier to build this bot when you have all those parts lying right in front of you, trust me!

Lets build our chassis

Gather together your tape player motor, your large value caps, and some heavy gauge wire. To build you chassis you’ll need to glue you caps onto the sides of the motor, just remember we need to put those caps in a parallel circuit. You should either glue the caps with all the positive ends facing out, or with all the negative ends facing out (see illustration).

Once you have all of those caps on there, you need to solder the post together (see illustration). The outside post probably won’t reach each other, so get that wire out. Make the wire into a small ring just big enough to fit around the posts, solder it into the ring. Then take and solder all those posts onto the ring (see illustration). The last step is to solder the inner post together, these should reach each other but if they don’t, use the wire again.

(click for bigger pic)

On to the SE

Lets define what exactly a solar engine is

Solar Engine – a circuit used to drive a motor which requires several tens of milliamps, with a solar panel that produces only a few milliamps.

See the illustration for the freeformed layout of the 1381J SE. Pretty simple. If you’ve been involved with BEAM before, this circuit should seem quite familiar! Keep in mind when you make this circuit that you want to have plenty of lead to solder to, but not to much that its a burden.

Combining

(click for bigger pic)

You have the SE and the chassis lying next to each other, right? From here on out its a piece of cake. First thing to do is to solder you positive and negative leads to the rings on the caps, once you have that take you motor (some motor might have wire leads, some have solder tabs, use precaution either way, even the best screw up motors!) and hook it up according to the drawing of the SE, that middle lead goes to the negative and then you can hook up the other end to the positive (you can reverse that so the motor positive goes to the middle lead, but I just happened to draw it that way). Those leads should be stiff enough to hold up that solar cell, but if they aren’t, you can use glue or anything to hold it down, just make sure it works first!

Finishing touches

(click for bigger pic)

It works, but doesn’t move to well, eh? This is where the heat shrink and washers come in. first off, take some heat shrink and put it on that motor shaft, chances are that it will slip off right away. To rectify that put on another layer. From there you can add as many layers as you see fit, just remember, to get that nice rounded finish, cut the inner most ones shorter than the outer ones so they shrink inward to round off the edges. Now your Quaret works, but goes in little circles. It’s time to talk about symet behavior.

Symets exhibit random behavior depending on their surroundings and the available light. The key to making a symet interesting to watch lies in the angle it leans at. If your symet leans at a 20 degree or more angle, chance are that its hard to tip and it just makes little circles. If you symet leans at a 18 degree or less angle it will travel almost in a straight line. If its too small an angle, it will tip ever trigger. So here’s the down low on getting it to travel straight. The symet should trigger and the edge its leaning on should cause it to turn, but if you decrease the leaning angle, the motor shaft is more perpendicular to the ground causing it to torque the symet the opposite direction of the turn, making it travel straight!

The washer can be glued to the bottoms of the caps (see picture) to rectify angle. use hot glue here, this way, if want to move them to adjust angle, take a heat shrink gun (or equivalent, a.k.a. something hot) and melt the glue so you can move the washer.

One more finishing touch is to add electrical tape around the edges of the solar panel, it help to protect you solar panel from nicks and cuts. Just remember, A for aesthetics, keep it pretty!

How does it work?

By now your Quaret should be up and running, work well? I hope so, cause i was pleased with mine. Now its your turn to experiment, Symets are so simple you can easily come up with your own plans. Hope you no longer refer to them as the brainless ‘bots, because a good symet can even display problem solving abilities, as well as brute strength.

KEEP ON BEAMing!!!

-Spencer

Editors note: This tutorial uses the voltage detector Solar Engine circuit. One thing I don’t like about this is that all these parts (electrical) aren’t readily available. Check out Ben Hitchcocks tutorial on how to build a Photovore! To find out more about robots similar to Symets, join the BEAM e-mail list, by sending an e-mail to beam-subscribe@yahoogroups.com

This year (2001), my parents have assigned me the task of doing a science fair project. After many many brainstorming ideas, I’ve finally settled on the idea of a project about maze solving robots and algorithms. I read Robot Science and Technology’s article about the C* algorithm, with just a little bit of confusion. After 3 readings I still don’t get it, so I decided I’d better start off simpler. I’ve also played around with Maze Bots, and read over their listings of algorithms. After much deliberation I finally decided on 3 different algorithms to do a project on:

1. Random solving

2. Left/Right wall following

3. A branching search pattern where you return to the last branch after a dead end.

Random Algorithm

This is by far the simplest way of solving a maze. Mind you know, I didn’t say best or shortest or fastest, but simplest. You simply have your robot run around making a random decision to turn or not when it encounters a opening to the left or right. The only problem with this, as I mentioned above, is that I will be slow, and there is a good possibility that the robot will not find the exit in the time allotted. I.E. Your robot could wander for hours always taking the wrong turns. Needless to say, it is probably well worth it to invest some programming time into a better algorithm if your looking for speed or accuracy.

Left/Right Algorithm

Ahhh … the amateur roboticists favorite. The whole principle behind this algorithm is that you can solve any continuous, i.e. no "islands", maze by following either the right hand or left hand wall. This will always get you out, unless the finish is is a "island," like the picture below.

In the above maze, a robot using the left/right wall following algorithm would never reach the exit. This algorithm is just slightly more complex to code, but it’s benefits over the random algorithm are large. Simply have your robot turn to the left (or right) whenever it encounters a doorway. Again, the downside of this algorithm is speed. One wall may continue for a long way before reaching the end.

Branch And Return Algorithm

Branch and return is simply a name that I made up. I’m sure there is some technical name, but for now that name will do. This is the most complex out of the three that I have chosen for my project. The principle behind this algorithm is that by exploring each branch of the maze you will eventually find the exit. This algorithm requires that you "remember" when you come to a branch, and begin to record your steps from that branch. After exploring that branch and you come to a dead end, you simple follow your path back to the original branch and take the next turn. This algorithm require much more coding, and some way of knowing your distance and direction, like wheel encoders, or maybe a accelerometer and compass. The problem with this algorithm is that the robot could fall into a endless loop. For instance suppose we had a maze that looked like this:

If the robot is heading from the top of the maze toward ‘a’ it then may decide to take a right and follow the corridor until it reaches ‘b’, it then might turn left and reach ‘a’ again, and then follow back to ‘b’, and never realize that it is going in a circle. One possible way to combat this is to have the robot take a random corridor when coming to an intersection. Giving the robot a degree of "forgetfulness", i.e. having it forget intersections encountered long ago, could prevent it from being caught in a very long loop.

Concrete – Actual implementation

For my science fair project, I plan on running the same robot through 2 or 3 different mazes for each algorithm, and recording it’s time. Personally, I tend to change my robots chassis every few months. The biggest reason for this is that I usually have built the chassis out of legos, and/or tape. This makes for quick ripping apart and rebuilding. I’ve finally settled on a design that I think I’ll keep for a while. I’ve built my latest robot chassis out of balsa wood. The main body is a 6 x 6 x 0.25 in (15 x 15 x 0.63 cm) flat square.



I’ve mounted two 4 battery packs on the back end of it, and centered two servers on either side. I secured two 6 x 3 inch balsa pieces together with wood glue and some reinforcing pieces of wood to form the main base. For each servo I built a small box on the bottom that just houses the box of the servo. For wheels, I’ve used two of the large lego wheels. They can be screwed very nicely to the servo shaft, and the have a excellent grip because of the rubber treaded tires.



In the front bottom of the base I’ve mounted on of Craig Maynards BOBIRD (Brains On Board Infrared Detectors) for central obstacle detection. At the moment I’m having a bit of trouble with the detector seeing the floor.



I also plan to place at least two more IRPD sensors on the top front for extra detection, and possible a few on the back for reverse detection.



For processing power I’m going to use a OOPic, and possible a Basic Stamp 2. The OOPic’s high level programming language makes it much easier than the Basic Stamp’s dumbed down language. For distance ranging (if I ever wanted to turn it into a firefighter) I will either mount a Sharp GPDU12 on a servo in the bots middle, for rotational ranging, or stationary at the front of the bot, for frontal ranging. Most likely I’ll also mount a breadboard on the bot for easy prototyping (and in my case permanent circuits!). Basically, my whole chassis is held together by Elmer’s and wood screws. Next month, I hope to be able to talk a bit about the code and hardware for navagating the maze.


Continue Reading …

Fred is my first ‘photopopper’ style bot. I had designed and built a similar bot before, but that one had three motors arranged radialy around a central axis. It didn’t work nearly as well as Fred does.

The secret to Fred’s success is the modified FLED solar engine circuit. I tried a FLED circuit, but the low-light performance was abysmal. I needed to put my 60W reading lamp 10 cm away from the solar panel before it would fire! This was clearly inadequate, so something had to be done. I tried putting in extra diodes, extra capacitors, extra resistors, and eventually settled on the design shown here.

Now Fred will pop away quite happily 40 cm from my desk lamp, which is the height that it normally sits when I’m reading. He will also pop all day when indoors, so long as a window is nearby to let in some diffuse daylight.

Specifications:

Voltages:
Switch-on: 2.4 Volts
Switch-off: 1.4, or 0.7 Volts, depending on what mood Fred is in.

Dimensions:
43 (length) x 27 (height) x 38 (width) mm.

Solar Cells:
One 33 x 24 mm Panasonic 5 Cell amorphous array (Part # BP-243318).
Storage Capacitors:
One NEC 0.033 Farad capacitor, with a 47uF cap in parallel to lower the internal resistance at switch-on.

Motors:
Two namiki pager motors, with extra padding on the shaft, directly resting on the ground.

Electronic components:
2 x BC337 NPN transistors
2 x BC327-25 PNP transistors
2 x Red Flashing LED (FLED)
2 x 3.3 K resistors
2 x 33 K resistors
2 x 4.7 uF capacitors

Performance:
Fred is quite an active bot compared to my other mobile bots. At 11:00 in the morning sun, he takes two seconds to recharge after a big step, and one second to recover from a small step. Sometimes he takes big steps, sometimes he takes small steps. When he takes a big step, the capacitor discharges all the way to around 0.8 Volts, but when taking a small step the cap discharges to only 1.5 or so volts. This results in interesting behavior. It seems that he takes small steps when directly facing the sun, and big steps when facing at 90 degrees or more. If Fred starts out facing into the sun then he will take lots of quick, little steps of maybe 45 degrees and follow the sun around the sky. But if you start with the bot facing away from the sun, or at 90 degrees, then he will take giant 180 degree strides, alternating sometimes with little steps to aim more towards the center of the light pool.

Construction:

Lets start with some mechanics. While your soldering iron is heating up, grab a pair of pliers and the paperclip.

You need to cut both the inside and the outside loops of the paperclip at the point where the inner loop bends. This will give you aroundd rectangle, the size of which is dictated by the size of your paperclip.

By the time you’ve gotten that puzzled out your soldering iron should be well and truly hot. Tin (put solder on) the top of both fuse clips and the outer bends of the paper clip. Make sure that you heat all metal pieces up till they are very hot – and don’t have the motors in the fuse clips when you do this! Attach the fuse clips to a fuse or something that doesn’t mind heat. The solder should flow on easily, and shouldn’t be able to be picked off later with a file.

Don’t leave the motors in the fuse clips for the next step. Use a couple of old fuses or pencils or something, but don’t use motors. You now need to put the paper clip on top of the solder blobs, and attach both fuse clips to the paperclip. This is harder than it sounds. Remember to orient the fuse clips at 45 degrees to the paperclip (so they are at 90 degrees to each other) and get the little tab on the fuse clip around the right way so that your motors will actually fit in the way they are intended (pointing away from each other, not towards each other).

This is what it should look like when you’re done:

(Note the amount of solder I used. This isn’t due to mechanical ineptitude… at least that’s my excuse and I’m sticking to it!)

Now wait until it’s all cooled down and stick your motors in. Check that the angles are vaguely symmetrical. If not, start again

Now attach the negative lead of the capacitor to the center of the top rail of the paperclip. The paperclip should be on the opposite side of the motors to the capacitor. Bend the capacitor lead around the paperclip, and belt it with some heat and solder. Make sure that the positive lead doesn’t touch the bottom rail of the paperclip. You are going to stick the solar panel on top of the cap and motors, so make sure that they create a flat surface to anchor the solar panel to.

Now it’s time to start on the hard part: the electronics.

Firstly note the differences between BCxx type and 2N39xx type. This tutorial is aimed at using BC style transistors. If you’re using 2N39xx transistors then the pinouts are different, so you’ll have to be very aware of how your transistors should go in.

Put the BC327’s into the body of the FRED, next to the capacitor, and connect up the emitters:

Now put the BC337’s into the FRED, connecting up their emitters as well:

If you are using 2N39xx transistors, then these diagrams are incorrect. Flip these diagrams horizontally, and they will be right.

*** Important ***

Check, check and recheck all the transistor connections. Go back to the transistor pinout diagram and the schematic, and make sure that all the right pins are connected to the right places. It’s easier to do it now than later, trust me.

If you don’t do it now, you will be doing it later.

Now connect the base of the BC337 to the collector of the BC327: Note: Thanks to the eagle-eyed Joel Hirtle for picking up the mistake that was in the last sentence!

You will now need to put the 3.3k resistors in, connected between the collector of the BC337, and the base of the BC327:

Here’s what it will actually look like from above:

Now lay the 33k resistors to the side of the 3.3k resistors and connect one end of the 33k to the back end of the 3.3k:

Now connect your 0.22 µF capacitor between the forward ends of both the resistors:

You now need to attach your FLEDs to the robot. Bend the FLEDs in the angle shown here: (the negative side of the FLED is at the top of this picture)

The negative lead of the FLED should now be attached to the blob of solder on the top of the robot that conects the negative lead of the capacitor to the paperclip.

The positive lead of the FLED is attached to the 33k resistor and 0.22 µF capacitor:

All that remains now is to attach the motor leads and the solar panel. To check the motor lead polarity, grab a 1.5 V battery, and connect it up to a motor. Figure out which way around gives you a forward kick. Do this for the other side as well. (one batch of pager motors went different ways – the only way to know is to do this check). Now, attach the lead of the motor that you had attached to the positive terminal of the AA to the positive lead on the capacitor. Attach the other lead of the motor to the back end of the 3.3k resistor on the side that the motor lives on.

This shot isn’t that clear, but you should be able to see the orange leads from the motors here. The black leads seem to be lost in the ether… Now for the final connections: the solar panel. Connect the positive lead of the solar panel to the positive lead on the capacitor, and the negative lead to the paperclip:

Now for the acid test: Hold your creation up to a light globe or the sun. If the motors fire, then well done! If not, then check all your connections again. This circuit is a bit chaotic in nature – if the ‘dark’ motor doesn’t fire the first time around then it might fire the second time around. Don’t give up hope the first time the ‘wrong’ motor fires. Remember: your circuit worked on the breadboard, so it should work when freeformed. You did breadboard it, didn’t you?

If you are having trouble getting your FRED engine to fire well, put a 1 uF cap across the motor terminals. This is a bit counterintuitive but for some reason this makes the motors turn on harder. Thanks to Kyle Simmons for this tip!

For the more adventurous, here’s the schemtic of the complete FRED popper, complete with touch sensors, and a balance pot to make sure that your little FRED goes in a straight line. The bits in red are there to balance out any differences in the FLEDs, and the blue parts are there so that you can add tactile sensors to your creation. Good luck!

Editors note: You can visit Ben’s robotics webpage here.