Skip to content

Latest commit

 

History

History
8 lines (6 loc) · 470 Bytes

README.md

File metadata and controls

8 lines (6 loc) · 470 Bytes

Sensor-Network

Sensor network problem from NIPS RL benchmarking 2005

The primary purpose of this code is to investigate efficient learning techniques for multi-robot coordination. The main program is accessed via test.cpp while the bulk of the domain is inside the SensorNetwork class.

Learning algorithms can be tested by editing the Policy class. Testing reward shaping methods will require additional functions within the Sensor class for decentralised learning.