Experiment plan
Does sharing a discovery help the team finish more work—and how many messages does it take?
That’s the first comparison we want to run. Swarm Lab is available to explore, but no comparative research results are published yet.
First comparison: the sharing rule
Run the same incoming tasks with sharing off, regular updates, and selective sharing. Keep the map, robots, task arrivals, and seed the same.
- Change
- When robots send information to their neighbors.
- Compare
- Work completed and message traffic, including late and lost messages.
- Keep
- The setup, code revision, seed, and run record so the comparison can be repeated.
- Status
- Planned. No reported outcome yet.
Later comparisons
- How task claims and temporary coordinators recover after a connection fails.
- Whether a learned policy improves on fixed task and communication rules.
- Radio delivery, energy, and computation on a physical Node.
Failed and inconclusive runs will stay in the record alongside successful ones. Simulator results won’t stand in for physical robot measurements.