Like its namesake, Stitch is small, compact, and aggressive. Each of its two wheels uses a gear train with a 25:1 ratio and powered by 2 motors. Thus, Stitch is quite a speed demon. There is an additional dummy wheel in the back, upon which the battery pack rests, and 2 wheels in the front to provide balance. The handyboard rests on top of the 4 motors. Legos are build around the wheels, gearboxes, and motors to strengthen the body of the robot. The CLAW is attached to the front of the robot. It has 2 degrees of freedom; one servo allows for an up/down motion of the CLAW and a second servo opens and closes the CLAW. Thus the ever versatile Stitch can grab a ball and lift it up and down.

Those are just the physical aspects of the robot. The beauty lies in the sensors. Two pairs of LEDs and phototransistors are placed at the bottom right and left of Stitch to help her determine what side of the table she is on. There are two roller switches on each side to help her wall-follow. The CLAW has a microlever switch and a small switch to help Stitch determine whether she has grasped a ball. Once Stitch has grabbed the ball, she uses LEDs and phototransistors (which are integrated into the CLAW) to tell whether the ball is white or dark green. If Stitch likes the ball, she will simply pick it up and travel. But if she doesn't...she'll blast it to smithereens with her laser gun! well...not quite. But Stitch does have the cool CHUCK move in which she rotates her claw more than 90 degrees upwards and releases the unwelcome ball so that it travels down the green ramp to its final (and worthless) destination.

Our teams goal was to not lose. In analyzing the rules of the contest, we observed that if one of 'our' balls is in the cup at the end of the round, then we score at least 4 points. The opposing robot can, in that case, score no more than 4 points, so by ending the round with the ball in the cup, we ensure a non-loss. Over the course of the contest, we discovered this was much more difficult than we originally realized. We modified our goal to consist of scoring one point and inconveniencing the other team by sending their balls to the bottom of the table.

Our robot first determines which side of the table it is on using LED/phototransitor pairs on its base. It then backs up and turns up the table. It uses wall following to reach the top of the ramp, then switches into ball retrieval mode. In this mode it will try to grap balls and then determine their color. If the ball is our own, Stitch will turn around and drive to the middle platform, using the LED/phototransistor pairs on the bottom to decide when to stop. It will then stay put and hold the ball. If it is a ball of the other color, Stitch will chuck it and continue searching for our color ball.

Previous two photos courtesy of Stan Hu and The Tech.

Experiment #626

Our initial goal was to be able to put a ball in the cup. The height of the cup edge necessitated the ability to lift the ball. We decided to do this using a claw.

Image from Disney

The initial claws that we experimented with had long arms to ensure that they would have the reach necessary to lift the ball high enough. From the beginning we recognized that two degrees of freedom would be necessary. In the first design, the servo in the middle of the claw moves two axles relative to each other, giving the claw the ability to lift. Another servo would have slid the arms along the rotating axles so that the claw could open and close. This claw design was eventually rejected for its complexity and bulkiness.

The second claw design was much more compact, and the use of gears to facilitate the opening and closing made that action more robust. With this design the servo was legoized which facilitated bracing of the mechanism. We continued to use axles to give length to the arms. As you can tell by the several different connectors used to attach the axles to the claw and servo, we had identified these connections as the weakest part of the mechanism, and we were experimenting to find the strongest one. The rubber bands were used on the claw here to increase the friction between the claw and the ball and also to increase the amount of contact area between claw and ball.

Here you can see how the second claw design fits around a contest ball. We've added another servo for a second degree of freedom. We were able to avoid a potentially weak connection there by attaching the gear on the second servo directly to the legoization mounting of the first servo.

In design three, we completely eliminated the axles as a supportive or structural element. In this design they are only used for bracing purposes. This increased the amount of stress the claw could handle. In addition, this design improves on the hand structures of previous claw designs. The cage-type design that you can see above was meant to improve the grasp that the claw had on a contest ball.

This is the final design of our claw. In the end, a simpler design prevailed. Due to the strength of the servos, the claw didn't have a problem gripping the ball, even without the use of rubber bands. The weakest part of our design continued to be the use of the lego axles, and in the first round of the final contest, the axle holding the black claw-half did pop out of the gear when our robot was attacked.

It was clear though that our claw had the needed functionality. It could pick up a contest ball successfully and reliably.