Let's assume you are building a LEGO Mindstorms or Vex IQ based Sumobot, but you want to skip some of the basic mistakes beginners will make. Here are some tips and tricks.
- Know the rules. It's silly to get disqualified because you didn't pay attention to the rules. Know the size and weight limits. Know the allowed construction materials and techniques. Know the startup behaviors. For example, your robot must not move for 5 seconds after you activate it. This simple rule has tripped up many competitors. And make sure you get to the competition on time to register and get your robot inspected for weight, size and any other requirements.
- Stay on the field. For this you will almost certainly need at least one light sensor to detect the ring's white edge. We highly recommend two such light sensors placed at the front corners of your robot. This will increase your chances of detecting the edge when coming at it from an angle. You can also adjust your retreat behavior so the robot will be less likely to exit the ring. A retreat behavior usually consists of backing up and turning back toward the center of the ring or scanning for your opponent. You can back up curving away from the sensor that detected the edge first. A third edge detector could be placed at the back of your robot - but this is almost never needed. It would be only useful if you have a behavior that could trigger backing up when the rear of your robot is close to the edge. Theoretically you could detect the edge when being pushed backward by your opponent and try to twist out of the way, but we've never witnessed anyone pulling off this advanced behavior.
- Build to the maximum weight for the competition. If your bot is heavier than your competitor's, you will have an advantage in traction and with inertia. You will be harder to push around and can more likely push them around. We've seen teams use extra unpowered motors to help maximize weight. Use a scale to be sure you don't exceed the allowed weight.
- Build compact. Your robot should be as small and dense as possible. Air gaps within your robot and on the exterior should be kept to a minimum. The larger your robot is, the more likely that your opponent will contact a part of your robot far from its center of mass. When it pushes against this part, it will very easily turn your robot in a different direction. Most likely this will be to your disadvantage. You will also be very unlikely to push your opponent in the correct direction when in this condition. Also, the rules say that the first robot to have any part touch the surface that the ring is sitting on is out. If you have a large robot, it is much more likely that part of it will touch-out.
- Build Low. The lower your center of gravity, the less likely your opponent will be able to topple you or force your wheels to lose traction.
- Build a Skirt or Shield. A Sumo Shield is a smooth ramp that decends from the front of your robot down to the surface of the ring. The purpose is to create a wedge that will go under your opponent when you come into contact. The wedge will lift your opponent, transferring their weight to your robot. As a result your wheels can increase traction while theirs will decrease. A skirt is a shield that surrounds your entire robot, making it look like a cone or pyramid, so it works wherever the contact point is. But a skirt can be much harder to engineer. They have to be very sturdy, not impede your own movement, and not get in the way of any sensors you might use. Skirts and Shields also increase the size of your robot, so you have more risk of touching-out. Particularly if you have a hinged shield. Hinged shields are great for staying as low as possible to get under your opponent, but they need to be prevented from dropping down when over the edge of the ring. A floating skirt is a wall built around your robot that is only loosely connected or not connected at all to your robot. Instead your robot pushes the skirt around the ring and the skirt's weight keeps it flat against the floor. This makes it unlikely that your robot's motions will create a gap that your opponent can get under. And if your opponent does get under the skirt, they haven't necessarily started lifting your robot to steal traction. You could also have a sensor that detects if your skirt is lifting and back away when that happens.
We've seen well-engineered robots with only edge sensors win big competitions. A solid, heavy and low robot with a great skirt will conquer when none of its opponents has the same features. Once you are in this category you can consider advanced tips.
- Locate your prey. Actively seeking your opponent creates an advantage. It's also fun. Usually a forward-facing ultrasonic sensor is a good choice. You can scan for your opponent by making your robot turn in place while checking the sensor to see if it detects something close. Calculate the maximum distance your opponent can be from your ultrasonic sensor. Simply place your robot backed up to the edge of the ring and measure the distance from the front of your ultrasonic sensor to the opposite edge of the ring. Subtract the minimum size of an opposing robot. For LEGO sumos that would be about 6 in. or 15 cm. If you see anything closer than this you can assume that you've detected your opponent. (Or you've detected humans if you've failed to keep everyone at a proper clearing distance from the ring, including the operators) Continue your turn for a fraction of a second and turn on your charging behavior. Make sure you are aware of the minimum distance your sensor can deal with. You will probably want to recess your sensor from the front of your robot so that it will continue to register your opponent even when you are right up against each other.
- Organize your software. Beginners will often design software that will do one thing at a time and be unable to react until those things are complete. For example, on detecting an opponent, charge for X rotations of the wheel. While the robot is trying to complete those rotations it's not looking at sensors, so it doesn't detect the ring and drives off if it was too close to the edge. We will post a complete lesson on designing software that always lets the highest priority behaviors (back-away-from-the-edge) interrupt the lower priority behaviors (scan-for-prey).