World Modeling

Purpose

The world modeling module processes that data from the vision system to provide filtering and prediction for the rest of the system. The world model will effectively decouple the vision system from the rest of the system by allowing any process that needs world data (robot and ball position and velocity) to get a predicted value at any time.

Robot Modeling

Each robot will be tracked with its

• Shell ID
• Position (X, Y)
• Angle (degrees)
• Current Velocity (X, Y, angle)
• Buffered States - maintain a circular buffer of previous corrected states. This will be used to calculate velocity and acceleration.

Ball Modeling

The ball should have its position (X,Y, angle) and velocity (X, Y, angle) tracked by filters/smoothers. The ball should handle multiple hypotheses to allow for handling of missing location of the ball.

Tasks

• COMPLETE - Integrate into module architecture
• IN PROGRESS - Implement basic Kalman filter for robot positioning (both with and without access to control)
• NOT STARTED - Implement Kalman filtering for ball positioning
• NOT STARTED - Investigate Smoothers
• NOT STARTED - Investigate particle filters (ConDensation)