This week my work was split up into three parts.

The first of these was the frame. I added a hole at the top shelf of the frame for the projector to sit on so that it could project the new image back on the table. We ran into a problem with this though as we didn’t realize that the projector projects its image on a tilt. If we stick with the current design, we only project on the top half of the pool table and not the bottom. This would require a new frame design and I came up with two main ideas. The first of these is to add a new leg to the third side of the frame as this would allow the projection to be over the entire table. The second is to extend the base and move the pool table forwards and this seems to be the easier solution.

The second of these was the shot calculation system. Until now I was drawing lines to the center of each ball, which was slightly incorrect as on an angle, the cue ball is not supposed to hit the ball in the center. So I corrected this by adding a ghost ball (where the cue ball is supposed to make contact with the target ball) and accounted for this when making the shot. This also required a redesign of collision calculation which was done since the current system was modular enough.

The Third was integration with Jimmy’s CV system. As of now, we are able to detect changes in the game state from the CV system, store these changes in a global values, and once we have these changes stabilized (i.e ball stops moving) we are able to calculate the best shot. Jimmy and I will be further testing this tomorrow and hopefully we don’t run into too many issues.

For the next week, My immediate priorities are going to be redesign the frame, fully integrate with Jimmy’s system and then start integrating with Justin’s pool cue stick system. I am on track with progress and we are doing well as a team at the moment.

This Week I made a lot of progress. I redesigned the entire software stack for the shot calculation since it was extremely buggy and slow. After a rewrite of the entire code base for calculation, today I removed the last bug from a preliminary shot calculator and as of now given a random game state, the software will select the easiest possible shot for the user and create an image for it. The main problem I addressed was the removal of certain pockets. Until now I was only removing the pockets based on the current position of the target ball, and now it has been updated to the maximum x and y coordinates the ball can go after its shot. A different bug I found was when the x coordinates for the balls were the same, the slope was coming out to be infinity which was messing up the entire system and now that has been accounted for. Moreover, with the rewrite of the code, the code is now a lot more readable and portable since I defined macros for a lot of the constants I use to minimize the number of changes that would need to be made in integration with the CV system from Jimmy. I am on track, and my next steps would be integrating my and Jimmy’s CV system and then mine and Justin’s cue system. After that is done, time permitting, I want to look into the option of adding bounce shots (off the railings) and chain shots (hitting one ball into another for the second one to go in).