Our research closes the loop between people and robots. We decode human state from neural and physiological signals, feed it into robot decision-making and learning, and evaluate the resulting behavior through human-subject experiments on perceived safety, trust, and team performance.
Identifying EEG, eye-tracking, and behavioral markers of how safe a person feels next to a collaborative robot, and how robot speed, proximity, and motion shape that judgment, as a basis for robots that adapt their behavior to keep perceived safety high.
Decoding control commands from EEG to fly drones in real time, as a testbed for studying neural control under time pressure and as a platform for hands-on student learning and outreach.
Design, prototyping, and evaluation of compact, lightweight drones for operation in confined indoor spaces, building on our work on micro-UAVs for first responders.
Developed at Lawrence Technological University with a multidisciplinary student team through the LTU Centrepolis Accelerator, this project designed, built, and evaluated a sub-250 g micro-UAV for emergency response, where small, agile aircraft can reach confined or hazardous spaces that responders cannot easily access. The work followed the full design cycle, from mission requirements and lightweight material selection to CAD design and prototype assembly, and validated the drone through structural analysis and FEA, stability, drop, and endurance flight tests, and RGB, time-of-flight, and thermal sensor tests. The final platform weighs 247 g, flies for about 28 minutes, and was built for under $3,000.
This NSF-funded project studied how trust is established during human-robot communication in law enforcement scenarios. We developed a teleoperated communication robot with audio, video, and display interfaces to support remote interaction between officers and the public, and examined how robot design and communication modalities influence trust in public safety contexts.
Using brain-computer interface methods, this project studied how people perceive and respond to different robot behaviors during collaborative tasks. EEG and behavioral data were collected while robot expressivity and gestures were varied, to understand how design features influence perceived safety, trust, and engagement.
A systematic review of human-robot collaboration systems that identified sources of uncertainty in collaborative tasks and analyzed how communication modalities affect trust and team performance. The project proposed a framework for purposeful human-robot communication to support more reliable and transparent collaboration.
This project explored how EEG signals can be used to decode human intention for adaptive interfaces. Machine learning models were developed to classify neural patterns associated with finger numeral configurations and event-related potentials, showing that these neural signatures can be reliably distinguished.
This project extended EEG-based intention decoding to closed-loop control for rehabilitation. Neural signals were integrated with a functional electrical stimulation (FES) cycling system to modulate cadence in real time based on user intent, showing how brain signals can guide adaptive rehabilitation devices.
