Purpose
We currently have a bespoke way of trusting or not trusting cameras that is based on the current robot and game. We should make this more modular and transferrable to future robots/games.
To that end, the current system is also brittle to cases where the untrusted cameras are actually reliable due to robot orientation. To generalize the implementation, we should determine based on robot heading and camera-measured position if the camera should be trusted. This can be done by computing if a tag is "in view" with some user-configurable viewing angle that is smaller than the camera FOV. If the tag(s) measured are not inside the trusted viewing window for a particular camera, we can reject the measurement outright. Otherwise it should be trusted just like any other measurement
Project Scope
[ ] Add calculation for if a tag is in view based on the robot orientation and measured location on the field
Purpose We currently have a bespoke way of trusting or not trusting cameras that is based on the current robot and game. We should make this more modular and transferrable to future robots/games.
To that end, the current system is also brittle to cases where the untrusted cameras are actually reliable due to robot orientation. To generalize the implementation, we should determine based on robot heading and camera-measured position if the camera should be trusted. This can be done by computing if a tag is "in view" with some user-configurable viewing angle that is smaller than the camera FOV. If the tag(s) measured are not inside the trusted viewing window for a particular camera, we can reject the measurement outright. Otherwise it should be trusted just like any other measurement
Project Scope