This website is under construction. More updates are on the way.

Sanjay OrugantiResearch Scientist · Rensselaer Polytechnic Institute

← Research

Research project · multi-robot systems

IKT-BT

Robots that learn by overhearing their teammates

Published in
IEEE Transactions on Cybernetics, 2025
My part
First author
Where
HeRo Lab and DICE Lab, University of Georgia
Builds on
KT-BT

Overview

Instead of every robot asking the same questions over and over, a robot can listen in on its teammates and learn passively, the way a person picks up a skill by overhearing someone explain it.

Robot teams usually share what they know by talking to each other directly. IKT-BT (indirect knowledge transfer through behavior trees) adds eavesdropping: a robot that overhears two teammates exchange knowledge keeps it too. Knowledge and control both live in behavior trees, so what is overheard can be used at once, and the team spreads procedural knowledge without crowding its communication channels.

How it works

Four knowledge-transfer modes between robots: query-response, query-response-update, eavesdrop-update, and eavesdrop-buffer-update.

Modality 1

Eavesdrop–update (EU)

A robot that overhears a query–response exchange between two teammates learns from it immediately, updating its own behavior tree with the knowledge it intercepted.

Modality 2

Eavesdrop–buffer–update (EBU)

Overheard messages are held in a buffer before they are used to update the behavior tree, an episodic-memory strategy suited to resource-limited swarm robots.

Results

Explore further

Cite

IEEE T-CYB

IKT-BT: Indirect knowledge transfer behavior tree framework for multirobot systems through communication eavesdropping

S. Oruganti, R. Parasuraman, and R. Pidaparti

IEEE Transactions on Cybernetics, 55(6), 2558–2570 (2025).