← projects / in progress Co-author

IROS 2026 Submission

“Systematic Hardware-Software Platform to Implement and Evaluate Multi-Robot Signal-Source Localization in an Indoor Environment,” submitted to IROS 2026. Built in the ADAMS Lab at the University at Buffalo.

What the paper shows

We built a hardware-software platform for systematically evaluating decentralized multi-robot signal-source localization indoors. Each robot maintains a local Gaussian-Process surrogate of the signal field and runs Bayesian Optimization locally to decide where to move; agents share posterior updates with neighbors over a noisy communication channel. The contribution is the integrated platform (hardware, comms stack, software architecture, and evaluation methodology) that lets the same algorithm be tested across RF, LoRa, and WiFi signal modalities without rewriting the experiment each time.

We validated with 4 physical TurtleBots in hardware and 50 agents in simulation.

My contribution

I led development of the decentralized swarm system: the distributed inference pipeline (per-agent GP + BO with online posterior sharing), the ROS 2 stack the agents run on, and hardware bring-up on the TurtleBots. I also contributed to the simulation harness used for the 50-agent runs.

Stack

ROS 2, Python, PyTorch, Gazebo for simulation, TurtleBot4 hardware. RF / LoRa / WiFi for the signal modalities.

  • Paper: submitted to IROS 2026; link will be posted on acceptance.