Investigate other ball models

[itsarune]

Description of the task

BallFilter uses a cached number of vision detections to estimate the ball position and velocity. Our implementation uses a simple linear regression to predict the ball position or velocity.

Other teams such as TIGERs and Zjunlict seem to calibrate their ball model at the start of every game on each field which seems to have a positive effect on their gameplay. Investigate alternate approaches from other teams' TDPs and see if we can improve our ball modelling.

Field testing is ideal, but we are able to simulate some vision drop-outs in simulation using ./tbots.py run thunderscope_main --enable_realism.

Acceptance criteria

• Research other ball filters and prototype them
• Evaluate whether these approaches are better by testing with the cameras
• Supplement [ball_filter_test.cpp](https://github.com/UBC-Thunderbots/Software/blob/master/src/software/sensor_fusion/filter/ball_filter_test.cpp) and ball_occlusion_test.cpp with the newer implementation

Blocked By

Related to

#3381