WinstonHartnett commented 5 years ago

The robot should be able to determine its relative position on the field based on vision and encoder tracking.

Specifics

Position is re-zeroed at the end of each trajectory execution, when the robot reaches a target. Approximate Relative Position (ARP) is tracked by estimating which "sector" (a division of the field based on target type and color). ARP is then used to determine which target the robot is facing when generating a trajectory, thus informing Relative Position (RP).

The robot uses a "sweep" to acquire a set of points ahead of it in case it needs to reset its position.

Potential questions to answer:

Is this feasible with encoder inaccuracy and vision template fitting?
Should LiDAR or sonar be considered?
How should collisions and/or obstacles be handled?

WinstonHartnett commented 5 years ago

RPLiDAR A2 sensor with Rust API: http://bucket.download.slamtec.com/351a5409ddfba077ad11ec5071e97ba5bf2c5d0a/LR002_SLAMTEC_rplidar_sdk_v1.0_en.pdf

WinstonHartnett commented 5 years ago

Rust API repo can be found here: https://github.com/cnwzhjs/rplidar.rs

FusionCorps / 2020-Green

Robot positioning system #3

Specifics