This is a collaborative project between Michael Mu, Romika Sairam, and Sachin George.
This project is an implementation of reinforcement learning to the field of swarm robotics. This means we wish to introduce a massive number of agents with only the simple ability of moving to work together to accomplish more complex tasks.
Currently, the goals of the project are as follows:
- Collaboratively search for a target.
- Collaboratively move a target that is too large for an individual agent to move.
- Coordinate to move the target to a destination.
Some terms we use are self-defined to increase simplicity of language. Their explicit meanings are listed here.
Target - A part of the environment which we wish to use our agents to find and move. In our default environment, this is a ragdoll humanoid.
Destination - A location in the environment which we wish the target be brought to.
The real-world equivalent of this environment would be an unconscious individual that we bring to a safe location or something similar.