Contents

Basic Robot-Building Tips

Your robot will need to move.  The contest has a 5 minute time limit.  If your robot can't go faster than 10 feet-per-minute, you may need to go back to the drawing board.  Most people use a two-motor design with two drive wheels and a caster.  You can play with the size of the wheels to see if more speed or more torque is better for your robot.  Try different sources to find the best parts.  Look in toy stores, hobby stores, hardware stores, and even sporting goods stores (in-line skate wheels). 

Your robot will need to have an easy-to-modify base on which to attach all of the components.  Ease of modification is essential since your robot will evolve over time.  The entire structure may not be larger than a 31 cm.  cube, but there is no weight limit.  You can use any material for the base, including plywood, machined aluminum, plastic, or Plexiglas.  Erector or Meccano sets are good too.  So are Legos.  If you don't want to build a robot from scratch, you may want to consider buying a kit, adding some sensors, writing some code and entering your robot in the new commercial division.  One note about robot bases...  A round robot usually works better than a rectangular one.  Rectangular robots can get jammed into corners or walls. 

Your robot will need a central computer or controller.  Popular choices include the Parallax BASIC Stamps, BasicX controllers, HandyBoard , OOPic, or even a laptop or Palm Pilot computer.  Our BrainStem controllers will also be there. 

Your robot will need motors and/or servos.  Prices can range from free (for that hackable toy you "borrowed" from your nephew) to hundreds of dollars.  Every robot would love Swiss-made high-efficiency motors with planetary gearboxes, but a surplus motor will also get you from point A to point B.  You will have to play with weight, position, power, and speed.  You can get fancy with multiple drive trains for X & Y travel, a tank tread design, or even a walker, but we advise saving time and money by going with the simplest system. 

Rechargeable batteries will save you money.  Over the time it takes to develop a robot, you will go through lots of batteries.  NiMH batteries are the most efficient.  Ni-Cads are good too.  Lead-acid batteries have great performance, but can be rather heavy.  Separate power supplies for the logic and motors can eliminate troublesome glitches. 

Advanced Robot-Building Tips

Mrs Stampy negotiates the maze on two feet!

Robots love the latest and greatest sensors.  It's all they talk about at robot parties.  Give your robot a Hamamatsu UVTron flame detector .  Its extremely reliable.  Give your robot some Devantech UltraSonic Rangers or Sharp GP2D02 (digital) or Sharp GP2D12 (analog) obstacle detectors.  Sometimes a homebrew solution is just as good as a commercial one.  But you must consider the time it takes to build your own vs.  the cost of buying an equivalent component. 

Every minute adding convenient switches and status indicators (LED, beeper) will save you an hour of grief later on.  It's nice if you don't have to turn your robot off by taking the batteries out.  Invest in a crimping tool, crimps, headers, and crimp housings.  Good connectors make it easier to maintain your robot.  If you use beep starting, add a manual backup switch and-or make the frequency easily tunable to someone else's beeper.  It's easy to misplace things (like beepers) at the contest. 

A multi-spectral approach for candle detection is quite effective.  Lighting conditions can easily spoof a system that relies solely on Infrared or visible light.  It's much harder to spoof a system that combines UV, visible, and IR sensors.  In addition to UV-IR sensors, the Eltec Pyroelectric sensor provides true heat-sensing capability.  However, it is a heat differential sensor, so it must sweep for best results.  The signal processing for its analog output can be as simple as a window comparator circuit. 

Use a fan if you want things to be simple and reliable.  Turn it on and blow out the candle.  The contest has been improved to offer points for more novel approaches so if that is where you want to spend your effort, consider an alternative.  Water, CO2, balloons, flour bombs, etc.  are nice, and often more satisfying than a fan, but when you use them up, you're toast.  You can blow a fan as many times as you want, repositioning the robot if necessary until you put the flame out.  Consider the benefits of a fan and then make your choice of what to use.  Perhaps your fan alternative can be reusable which would both give you bonus points, and keep things fault-tolerant. 

Testing Tips

Do not underestimate the horrifically harsh lighting conditions at the contest: blinding gym lights, moving shadows, candle reflections on bright white walls, flash bulbs reflecting on black floors, IR from camcorders and cameras, etc.  You will use light for nearly everything---candle detection, obstacle detection, line detection.  To test, turn on every light in your testing area, and then some.  Try pointing your TV remote at your robot.  Take some flash pictures while its moving.  Bring in bright flashlights and make shadows.  If your robot passes these tests and laughs at your futile attempts to fool him or her, then you have a chance at the contest.  (For the 2001 contest, there was an attempt to make the lighting more uniform, but it had an unintended side effect--- lots of UV noise.  So it is wise to plan for the worst.)

Some corners may have highly non-reflective rubber joints that can spoof a reflection-based IR obstacle detection system.  Add that to your testing area---black electrical tape might work.  See if your robot can handle it. 

An empty round oatmeal container or a piece of PVC pipe filled with pennies, rocks, or sand makes a good substitute for furniture.  The contest furniture is made out of heavy steel pipe.  Each tube is 11.5cm in diameter and weighs about 5 pounds.  The diameter is about the same as a round oatmeal container.  Some "safety yellow" spray paint will give you a good color match. 

If you are using a compass to help your robot navigate, be aware that the magnetic field can be distorted by nearby pipes and other metal objects.  In fact, compass readings can vary noticeably from one arena to the next at the contest site in Hartford.  If you are running in furniture mode with a compass, the steel pipe of the regulation furniture pieces may cause inaccurate compass readings.  You may need to use metal furniture cylinders for adequate testing.  A fencing or construction materials supplier may have the steel pipe you need to make your own furniture.  Ask them to cut it for you.  (This compass tip is courtesy of Ken Boone.)

Nancy avoids furniture at the Acroname 2001 Robot Expo.

Exclusive use of dead-reckoning (measuring the course and programming your robot to follow the measurements) for navigation will require you to make a full replica of the house for serious testing.  If your navigation system does not rely on dead-reckoning, you can test in small pieces of the house.  This is important for those of us with space limitations (like small apartments). 

Test your candle detection and candle extinguishing systems at the minimum and maximum possible heights of the candle (15 to 20 cm.  off the floor).  Test them with your robot as close to the candle as it can get.  Test them at the maximum legal extinguishing range (30cm). 

As batteries are used up, motors slow down and robot timing changes.  In the Expert Division, the floors in the rooms have different textures.  Your robot will probably move slower on carpet than on linoleum.  If your robot uses motions that are based purely on timing, you will need to test it with different power levels and floor surfaces. 

General Hints from Judging an Arena

Keep it simple.  Many of the most successful robots rely on a simple design, simple software solution, and straightforward mechanics.  A basic design that can navigate the maze, find the candle reliably and put it out will always score well if it can succeed three tries in a row. 

Expect changes to the environment.  Almost all problems are followed by the entrant saying "that doesn't happen in my lab/school/basement." The most common problems are changes in lighting, including cameras, or other forms of interference.  Try your church or local high school gym to test the lighting.  Use several video cameras to film your robot to see if the camera's IR rangers cause problems.  Also, trip a strobe or flashbulb to see if it confuses your robot.  The mazes are all slightly different with different floors, wall joints, and lighting.  Your design will need to be robust. 

Make as few changes at the contest as possible.  Don't try to rewire your robot or program it at the contest.  This takes planning ahead of time but it pays off.  Last minute changes are probably even less helpful than late night cramming before an exam.  Simple changes you may want to try are shrouds for masking your sensors from interference, threshold settings based on lighting, and slight variations of sensor position.  These are all tuning and not redesign modifications.  Stick to what you brought and what you worked so hard on to get ready. 

Watch other robots perform and visit the other builders.  Often, you are so focused on getting your robot to work that you don't learn from the other designs and people around you.  This is the real benefit from the event so take advantage of it!

One More Thing You Should Do at the Trinity Contest

Attend the talks presented by the guest speakers!

Contributors and Rules

Mark Whitney, Trinity contestant in 1995, 1996, 1999, 2000, and 2001.  (Senior Division 3rd Place in 1996, Special Recognition in 1999 for Stampy, Most Unique Robot Design in 2000 for Mrs.  Stampy, Expert Division 3rd Place in 2001.) He learned most of this stuff the hard way. 

Steve Richards, Trinity judge in 1999, Guest Speaker in 2000, and representative of Acroname in 2001. 

For a history of the contest, specifications for robots, and rules for competition, please refer to the contest rules and arena specs at Trinity Contest Rules and the contest pages. 

Our thanks to Jake Mendelssohn and staff at Trinity College for creating and continuing to run a first-rate robotics competition. 

If you find this article interesting and would like to be notified of future changes or new articles, feel free to sign up on our mailing list. 

If you have further information you would like to add to this article, feel free to email us with your thoughts as well as some information about yourself.  If the information is appropriate, we will add it and site you as a contributor. 

Revision History:

• 2006-11-06: Changed to conform to the non-fan bonus. Thanks to Ericson Ace Mar for pointing this out!

Related Links:

Using A Window Comparator When An Analog Input Is Not Available

Acroname 2001 Robotics Expo