Ramviyas Nattamai Parasuraman
       ASSISTANT PROFESSOR, UGA CS

Multi-point Rendezvous and Multi-Robot Herding: In this work, we introduce a coordinate-free multi-point rendezvous control strategy to enable multiple robots to gather at different locations by tracking their hierarchy in a connected interaction graph. A key novelty in this strategy is the gathering of robots in different groups rather than at a single consensus point, motivated by autonomous multi-point recharging and flocking control problems.

Multi-Robot Hierarchical Rendezvous Control: Using a hierarchical network topology tree, we devised a rendezvous control algorithm that can work on distributed fashion using the local sensing information at the robots, to collect all the robots rendezvous at a desired location in a hierarchical movement aided by the bearing-based source tracking of radio signals.

Resilient Communication-Aware Motion Planner (RCAMP) (with Sergio Caccamo): We developed a resilient motion planner for mobile robots to find an optimal path that guarantees traversability and wireless connectivity. The RCAMP can also be used to re-establish the wireless connection in case of a communication loss with a mobile robot.

Human-Robot Interface (HRI) for Wireless Network Connectivity Situation Awareness in UGV Teleoperation (with Sergio Caccamo): We extended a UGV Teleoperation Interface called Free Look Control (FLC) with the Wireless Network Connectivity Information. We used directional antennas to obtain the Direction of Arrival (DOA) (or gradients) of the Wi-Fi Received Signal Strength. Using this DoA we devised a Human-Robot Interface (HRI) that can support a teleoperator with Situation Awareness (SA) on wireless connectivity.

Wisture - A Solution for Gesture detection in smartphones using WiFi signals (with Mohamed Haseeb): Wisture is a new online machine learning solution for recognizing touch-less dynamic hand gestures on a smartphone. Wisture relies on the standard Wi-Fi Received Signal Strength (RSS) using a Long Short-Term Memory (LSTM) Recurrent Neural Network (RNN), thresholding filters and an artificial traffic induction approach. Look at the Codes and Datasets page for more information on the Wisture.

Gaussian Processes based Radio Signal Strength Prediction on Mobile Robots: We used Gaussian Process Regression to predict spatial changes in the RSS on the path of a mobile robot. The solution is composed of two stages. First, the location of radio signal source is predicted using a grid-based RSS gradient estimation method. Second, an online iterative Gaussian Process Regression (GPR) solution is used to predict the RSS in the future robot positions with reference to the current position of the robot.

Reinforcement Learning of Behavior Trees using Genetic Programming (with Michele Collendanchise): Taking advantage of modularity and reactiveness of Behavior Trees, we propose a model-free Automated Planned framework using Genetic Programming to derive an action plan for an autonomous agent to achieve a given goal in unknown environments. The advantages of the proposed approach are based on the advantages of BT over Finite State Machines. In particular, our approach avoids the problem of logic violation during the learning process.