This project develops a multi-agent robotic platform to study human-robot teaming during monitoring missions. The platform enables experiments on trust, decision-making, and performance when humans teleoperate robots alongside autonomous agents. For example, it investigates how trust changes when autonomous robots malfunction and human operators must adapt.
The system uses multiple robots in a controlled lab environment where monitoring tasks are represented by QR codes distributed across the space. Robots—some teleoperated by humans and others operating autonomously with random-walk strategies—collect these QR codes within set timeframes. A live overhead tracking system monitors robot positions, while onboard perception systems identify QR codes. Machine learning techniques are integrated to enhance QR code detection, robot navigation, and to model patterns of human-robot interaction.
-
Teleoperation of multiple robots in a shared lab environment.
-
Development and testing of autonomous random-walk strategies.
-
QR code detection and recognition using vision libraries and machine learning.
-
Integration of live tracking and behavioral modeling for experiments in human-robot teaming.
The platform is deployed in a robotics laboratory with a 9 × 18 m space and ten overhead cameras providing continuous tracking. Human teleoperators control two robots remotely, while three autonomous robots navigate and attempt to scan as many QR codes as possible in a 10-minute mission. Tracking data is stored and analyzed to study interactions between humans and autonomous agents.
Machine learning is applied to improve perception, automate decision-making, and analyze human-robot collaboration outcomes. This creates a flexible experimental testbed for exploring trust, performance, and adaptive behaviors in heterogeneous teams of humans and robots.