SAC (2015): Sumo Robot Competition
Introduction. On April 11th at Ohio State University, a team consisting of myself, Tal Singer, and Dmitri Tserkovnyuk represented Temple University in the Sumo Robot Competition of the Region 2 Student Activity Conference (SAC) 2015. Under the constraints specified by the rules of the competition, the team was tasked with the construction of a “sumo-wrestling” robot capable of pushing its opponent out of a dohyō -styled arena. In spite of the many challenges the team faced, our robot, unofficially named “Jawn”, managed to come in second out of twelve teams!
Video of probably one of the coolest matches of the whole competition. Be sure to check out my YouTube for more videos on the competition!
Background. Each of the matches themselves involved up to three rounds during which the robots of two opposing teams would attempt to push the opposing robot out of the arena to the point where a part of the opposing robot was touching the ground. The winning team of each match was the team who had won two rounds.
Another video of our team beating another opposing team.
The competition itself was a double knock-out tournament. Teams knocked out the first time were placed in a losing bracket. Teams knocked out a second time were eliminated from the competition altogether. The winner of the competition itself was decided on a match between the winner of the winning bracket and the winner of the losing bracket; however, if the winner of the winning bracket loses once, they got one last match with the winner of the losing bracket.
Image of the two arenas at the competition.
Developing and then constructing a robot according to the design constraints required by the rules was probably the most difficult aspect of the project. The weight of the robot needed to be no greater than 500 grams. To make matters worse, the robot itself had to be designed to fit within a 10 by 10 cm square on the ground. There was no specified height, but who would want to build a sumo robot with high center of mass? All matches took place on two different arenas, both of which were circular with a white line around the perimeter and a black center. The primary, larger arena had a perimeter width of 2.5 cm and a diameter of 77 cm. The secondary, smaller arena was only used for matches that made it to three rounds.
SolidWork pictures of the final design of the robot.
Design. The overall appearance and design underwent many changes throughout the duration of the project, starting with some very lofty ideas but eventually humbling down to the design shown in the figures above. One of the ideas the team conceived was to develop a pyramid-shaped robot with a spiral top. The idea behind this early design was to build a chassis that could act as the ultimate defense and offense, causing any opposing robot to climb upon the walls of the pyramid and then flip over due to the spiral top. I regret not getting the SolidWorks picture of this early design, simply because it looked extremely cool. Fortunately, the design on which the team decided was far more practical when it came to fitting all the components within the specified amount of space. Due to not properly accounting for wires, a hole still needed to be cut into the top of the 3D-printed chassis by the end of the robot’s construction, though.
The tank wheels and treads.
One early idea that remained throughout the project was to have tank-like treads and wheels for movement. Several ideas were considered on how to construct the tank-wheels. However, in the end, the tank-wheels were simply purchased and the motors were adjusted accordingly. The two DC motors were chosen based on balancing speed and power.
The embedded controller attached to the soldered prototype board.
The electrical aspect of the project was only challenging due to, once again, the limited space available. With the goal of minimizing the form factor of all electronic components, a prototype board was soldered together and connected directly on top of our embedded controller with headers, as shown in the figure above. The prototype board held smaller electronics such as regulators and the H-bridge needed to driver the DC motors. The prototype board/controller was situated above the robot’s batteries, tank-wheels, and motors.
The "guts" of the robot, including the embedded controller / prototype board, wheels / treads, and all the sensors.
The eyes and ears of the robot included two “proximity” IR sensor modules for determining at close range where the opponent is, two “ground” IR sensor modules for checking whether or not the robot is touching the arena’s perimeter, and an ultrasonic range module for checking if the opponent is directly in front at long range. The main reason for choosing the aforementioned components is that we already had them, and they worked well for what the team was trying to accomplish.
Tal's dog Snoopy played a critical role in keeping the team's motivation and mood high, especially during those late nights when we were trying our best to accustom ourselves to the sweet aroma of solder.
Finally, the embedded software was written in the C programming language to carry out the following algorithm. The algorithm was broken down into several “priorities”, to emphasize that certain conditions need to take precedence over other conditions.
Is the robot touching the perimeter? If yes, go the opposite direction!
Is the opponent close enough to and not directly in front of the robot? If yes, turn to face it!
Is the robot facing the opponent? If yes, go forward and attack!
Is the opponent detected from afar? If yes, go forward!
Is the robot moving forward for too long? If yes, start turning in one direction!
Is the robot turning in one direction for too long? If yes, start turning in the opposite direction!
The following is a summary of all the major components composing the robot.
Parallax Quickstart Board: Development board containing the embedded controller, a Propeller MCU.
Nitecore Lithium Ion Batteries: Power source.
Polulu Micro Metal Great High Performance Motor: Move and turn with power and speed.
Polulu Tank wheels and treads: Have large surface area to the ground while moving.
Parallax Ultrasonic Range Finder: See objects from afar.
Vishay IR Sensors: Detect objects up close so as to determine which direction to turn, and detect color between the arena’s perimeter and center.
From left to right, we have Dmitri, Andrew, and Tal. The reason for my ragged appearance compared to my teammates was that we took the picture during their senior design review. I'm in regular school attire, which I suppose is fairly ragged.
Team. Here are the lovely members of the team!
Major: Electrical Engineering undergraduate (Now graduated)
Role: Team lead (participated in all aspects of the project, kept team on schedule, etc.)
Role: Responsible for the electronics (soldering prototype board, integrating components into chassis, etc.).
Dmitri Tserkovnyuk
Major: Mechanical Engineering undergraduate (Now graduated)
Role: Responsible for the hardware (designed chassis, decided on wheels and motors, etc.)
Major: Engineering PhD graduate
Role: Responsible for the software (implemented algorithm and drivers, performed debugging, etc.)
More pictures of the team, doing stuff!
Results / Final Thoughts. And the result of the team’s efforts? We managed to get second in the competition! Personally, I was a little sad that our spiral-topped pyramid of amazingness did not come to fruition, however placing very well in the competition itself made up for such dramatic changes to our original design. Everyone in the team worked well with each other and managed to complete their tasks in a timely manner, which is excellent considering the date of the conference was near the end of the semester. I thought the whole experience was fun, and it was nice break from the graduate school life. Most of the research-related work I do tends to involve only me and my adviser, and the graduate courses I take don’t seem to involve any group projects, so I enjoy a good collaborative effort.
The rewards banquet!
It should be noted Dmitri didn’t attend the conference himself since he wasn’t an IEEE member. The whole team was mentioned in an article published on the college’s website.