LightR

From Robowiki
Jump to navigation Jump to search
Sub-pages:
Version History

LightR is a full rewrite of ScalarR, dedicated in 1v1. For more information, see Version History.


Design principle
Strategy light, machine learning heavy
Short-term goals
* Apply gradient descent into targeting & surfing models
* Rework energy management sophisticatedly
Long-term goals
* Apply reinforcement learning, the holy grail of machine learning