RESEARCH
JOURNAL PAPERS

 

"Adapting robot's explanation for failures based on observed human behaviour in human-robot collaboration",
IEEE International Conference on Intelligent Robots and Systems (IROS), 2025

This work develops a data-driven predictor that interprets human behavioral signals (facial emotions, gaze, gestures) from a user study and corresponding REFLEX dataset with 55 participants to anticipate confusion during robot failure explanations. The approach analyzes human reactions across failure, explanation, and resolution phases to determine if reducing explanation detail would induce confusion, then adapts the robot's explanation level accordingly for more efficient collaboration. Promising preliminary results show this mechanism can effectively minimize confusion while optimizing explanation verbosity based on real-time human feedback.

"REFLEX Dataset: A Multimodal Dataset of Human Reactions to Robot Failures and Explanations",
ACM/IEEE International Conference on Human-Robot Interaction (HRI) | 2025
🏆Honorable Mention Best Paper (Short Contribution), Best Paper Award Nominee

This work presents a comprehensive multimodal dataset, called REFLEX, capturing human behavioral responses (facial emotions, gaze patterns, gestures, timing) to robot task failures and subsequent explanations in collaborative scenarios. Recorded across failure occurrence, explanation delivery, and recovery phases, REFLEX enables analysis of how humans process robotic errors and interpret adaptive explanations, supporting development of context-aware failure communication systems. Awarded Honorable Mention Best Paper at ACM/IEEE HRI 2025, the dataset provides critical ground truth for training models that predict user confusion and optimize robot explanation strategies in real-time HRI

"YCB-Handovers Dataset: Analyzing Object Weight Impact on Human Handovers to Adapt Robotic Handover Motion",
IEEE International Conference on Robot and Human Interactive Communication (RO-MAN) | 2025

This work introduces a dataset of 2,771 motion-captured human–human handovers using YCB objects across varying weights, addressing a gap in weight-sensitive handover research for intuitive robotic motion planning. The dataset reveals how object mass systematically influences human reaching trajectories, timing, and readiness cues, enabling data-driven models for adaptive robotic behaviors that mimic natural handover patterns. Covering a wide weight range including objects requiring careful handovers, it supports robust analysis of weight effects on handover dynamics and human-inspired robot control strategies.

"Adapting Robotic Explanations for Robotic Failures in Human Robot Collaboration",
ACM/IEEE International Conference on Human-Robot Interaction (HRI), 2025. (HRI-Pioneers Abstract) | 2025

This work explores strategies for robots to dynamically adjust failure explanations based on human responses during collaborative tasks. The work analyzes user reactions to robotic failures and develops adaptive mechanisms that tailor explanation verbosity and content to individual user needs, improving task resolution efficiency and trust. Presented at the HRI 2025 Pioneers workshop, it demonstrates how real-time behavioral observation enables more intuitive and effective human-robot communication during failure scenarios.

"How do Humans take an Object from a Robot: Behavior changes observed in a User Study",
International Conference on Human-Agent Interaction (HAI) | 2023
🏆Best Poster Award

This examines how humans adapt their taking behaviors when receiving objects from robots in repeated interaction for robot-to-human handovers. Awarded Best Poster at HAI 2023, these findings provide essential insights for designing reciprocal robot handover behaviors that could better accommodate natural human expectations and movements.

"A Multimodal Data Set of Human Handovers with Design Implications for Human-Robot Handovers",
IEEE International Conference on Robot and Human Interactive Communication (RO-MAN)  | 2023

This paper presents a comprehensive dataset capturing synchronized motion capture (MoCap) and force/torque from human–human handover interactions, providing rich ground truth for analyzing natural handover kinematics and dynamics. The dataset reveals key patterns in grip release timing, interaction forces, and interpersonal coordination in human handovers that directly inform robot handover design, enabling more fluid and intuitive human–robot physical collaboration. Through detailed analysis of these multimodal signals, the work identifies actionable design principles for robots to better match human expectations in collaborative manipulation tasks.

"Detecting the Intention of Object Handover in Human-Robot Collaborations: An EEG Study",
IEEE International Conference on Robot and Human Interactive Communication (RO-MAN)  | 2023

This work investigates neural correlates of handover intention using electroencephalography (EEG) during human-robot collaborative tasks. The study analyzes brain signals to distinguish between handover-intended and non-handover actions, demonstrating that EEG patterns can provide early cues for robots to anticipate human intent before overt motion cues become available. These findings enable proactive robot behaviors that improve timing and safety in physical human-robot collaboration by leveraging brain-computer interface technology for intention prediction

"Effects of Explanation Strategies to Resolve Failures in Human-Robot Collaboration",
IEEE International Conference on Robot and Human Interactive Communication (RO-MAN)  | 2023

This work examines how different robot explanation approaches impact user trust recovery and task resolution after robotic failures in collaborative settings. Through a controlled user study, the work compares various explanation levels (high, medium, and low) and explanation variation strategies (fixed or decaying level) to identify which approaches most effectively restore human confidence and enable successful failure recovery.

"Human inspired grip-release technique for robot-human handovers",
IEEE-RAS International Conference on Humanoid Robots (Humanoids) | 2022

​This work proposes a data-driven grip-release strategy for robots that mimics natural human grip-release patterns observed in a human-human handover dataset. The method analyzes a human giver's grip release timing and observed interaction forces to create data-driven robotic grip-release policies that use interaction forces in robot-to-human handovers to coordinate object transfer more safely and human-like with human takers. Validated in real robot experiments, this human-inspired approach improves grip release timing and handover naturalness in robot-to-human handovers.

"A bio-inspired 3-DOF light-weight manipulator with tensegrity X-joints",
IEEE International Conference on Robotics and Automation (ICRA). | 2020

This work presents the design, fabrication, and control of a novel tensegrity-based robotic manipulator inspired by bird neck anatomy. The work develops a single-module prototype using tensegrity X-joints that achieves 3 degrees of freedom with significantly reduced weight compared to traditional rigid manipulators, enabling high dexterity in compact form factors. Through kinematic modeling, simulation, and experimental validation, the paper demonstrates the manipulator's potential for bio-mimetic robotics applications requiring flexibility and lightweight construction.

"Design Analysis and Development of Low Cost Underactuated Robotic Hand",
IEEE International Conference on Robotics and Biomimetics (ROBIO) | 2016

This presents the design, modeling, fabrication, and control of an affordable five-fingered robotic hand using underactuated mechanisms and proximity sensor-triggered actuation. The hand achieves multiple grasp types (power, precision, tripod, hook) through a single cable-driven system that adapts finger closure based on object geometry, significantly reducing complexity and cost compared to fully actuated designs. Demonstrated with forward kinematics validation and real-world grasping experiments, this work provides practical engineering solutions for accessible robotic manipulation in resource-constrained environments.