Paper accepted at Robotics and Autonomous Systems!
Excited to share that our paper, "Interpretable Responsibility Sharing as a Heuristic for Task and Motion Planning" has been accepted for publication in the Special Issue "Planning and Learning for Autonomous Robotics" of Robotics and Autonomous Systems.

In this work, we introduce Interpretable Responsibility Sharing (IRS), a heuristic for task and motion planning that helps robots make more efficient and explainable decisions by leveraging human-centric environmental bias. By reasoning about when to use auxiliary objects such as trays and pitchers, our approach improves planning effectiveness while maintaining interpretability through rule-based decision-making.

Paper accepted at Pattern Recognition!
Excited to share that our paper, "Locally Adaptive One-Class Classifier Fusion with Dynamic lp-Norm Constraints for Robust Anomaly Detection" has been accepted for publication in Pattern Recognition.

In this work, we introduce a novel ensemble anomaly detection framework that dynamically adapts classifier fusion weights using localized lp-norm constraints, enabling improved robustness to data imbalance and distributional shifts. This approach achieves significant computational gains with an interior-point optimization method and excels across both benchmark datasets and real-world temporal sequences.

A key highlight of this work is the introduction of LiRAnomaly, a novel robotics anomaly detection dataset based on pick-and-place tasks using a Franka EMIKA robot. The dataset captures both static and temporal anomalies—from sensor occlusions to trajectory obstructions—making it a valuable benchmark for safety-critical robotic manipulation.

Paper accepted at IEEE RA-L!
Excited to share that our paper, "H-MaP: An Iterative and Hybrid Sequential Manipulation Planner" has been accepted for publication in IEEE Robotics and Automation Letters (RA-L).

In this work, we present a hybrid approach that reduces configuration space dimensionality by decoupling object trajectory planning from manipulation planning, enabling robots to handle complex tasks involving tool use and bimanual coordination.