The ball-return project started with a mind map of what I wanted the system to be able to do and multiple ways that I could build the different parts of the project. The mind map helped me see that there were three basic parts to the problem: (1) capturing the ball (makes and misses), (2) determining the player’s location on the court, and (3) moving the ball from the capture point to the player.

I considered a wide variety of approaches, including not catching the ball but building a system to direct it back to the player from wherever it happened to come to rest -- for example, by having a robot fetch balls, or having a special floor that tilted or blew (imagine something like an air-hockey table) the ball back to the player.  But I decided that catching the ball in a large net and throwing back to the player was the most practical.

I explored three different methods for returning the ball to the player and built prototypes of each. The first way was with 3 hammers put together with zip ties and a long piece of metal to allow for a big swing with a lot of momentum. The 2nd way was with long 2x4 with springs attached to catapult the ball to the player. The last way was with 2 wheels connected to motors. From each of these 3 ways to launch the ball back to the player main things I needed the ball to be able to reach the player at the 3 point line (19.75 feet) with a good crisp pass at chest height. I found that the hammers were able to get the ball to a decent distance but to reach the 3 point line the ball would have to take one bounce. Another problem with using the hammer was having the hammer reset after it had swung to hit and launch the ball. Also, to be able to get the ball far enough the hammers and the metal rod would have to be very long  and ultimately make it very bulky and heavy. The second system of a spring-based catapult was able to throw a good distance but would also have to bounce to get all the way to the three point line. In addition from the catapult being spring loaded it would also have to be reset after each time is shot the ball to the player. The last way that I tried was with two motors that were connected to rubber wheels and an adjustable ramp system. This was by far the best method, with the ball easily able to get to the distance I needed it to and being the least bulky and very simple that would not involve having to be reset after each time it shot the ball to the player.

The last main problem I needed to solve was being able to know the players location of the court and have the ball get to the spot. The first idea was having weight sensors in the floor that would communicate to another machine of where the player is on the court. The second idea was using a camera something like what the X-box and the Wii use. The final idea was to have to player wear a device that sends a signal to another device that rotates to always face the player because of the device they are wearing. Conveniently I was able to find a machine called a Swivl that uses infrared and radio frequencies. The Swivl is composed to two parts a portable signal emitter and a stationary receiver. The Swivl ultimately solved this problem and with easy access to it I decided to use it.