Selected work

Representative Papers

Dull, Dirty, Dangerous: Understanding the Past, Present, and Future of a Key Motivation for Robotics

Feb 04, 2026

The robotics field has long invoked “dull, dirty, and dangerous” (DDD) tasks as a primary motivation for technological deployment, yet it has largely neglected to define DDD explicitly or provide concrete task examples, thereby hindering deeper understanding of human–robot labor dynamics. This study systematically examines robotics literature from 1980 to 2024 that references DDD, integrating empirical insights from social science on DDD work through bibliometric analysis and interdisciplinary review. We construct a labor-context-oriented framework for evaluating robotic technologies and find that only 2.7% of surveyed papers define DDD, and merely 8.7% offer specific task instances. The proposed framework offers the robotics community a novel perspective for assessing the socio-technical implications of automation, fostering more responsible and socially grounded technological development.

0 citationsRead paper

Assistive Decision-Making for Right of Way Navigation at Uncontrolled Intersections

Sep 22, 2025

To address the high accident risk at unsignalized intersections—stemming from ambiguous right-of-way, occluded visibility, and unpredictable driver behavior—this paper proposes the first POMDP-based decision-support framework for human-driven vehicles. Methodologically, we develop a high-fidelity simulation platform incorporating stochastic traffic flow, pedestrian dynamics, visual occlusions, and adversarial scenarios, and systematically evaluate three probabilistic planners—QMDP, POMCP, and DESPOT—against a deterministic finite-state machine baseline. Our key contribution is the novel application of POMDPs to right-of-way assistance for human drivers, revealing the critical role of explicit uncertainty modeling in safety-critical decision-making. Experimental results demonstrate that probabilistic planners achieve up to 97.5% collision-free intersection traversal; among them, POMCP attains the highest safety performance, while DESPOT achieves the best trade-off between computational efficiency and real-time feasibility.

0 citationsRead paper

Recent publications

Latest Papers

Dull, Dirty, Dangerous: Understanding the Past, Present, and Future of a Key Motivation for Robotics

Feb 04, 2026

The robotics field has long invoked “dull, dirty, and dangerous” (DDD) tasks as a primary motivation for technological deployment, yet it has largely neglected to define DDD explicitly or provide concrete task examples, thereby hindering deeper understanding of human–robot labor dynamics. This study systematically examines robotics literature from 1980 to 2024 that references DDD, integrating empirical insights from social science on DDD work through bibliometric analysis and interdisciplinary review. We construct a labor-context-oriented framework for evaluating robotic technologies and find that only 2.7% of surveyed papers define DDD, and merely 8.7% offer specific task instances. The proposed framework offers the robotics community a novel perspective for assessing the socio-technical implications of automation, fostering more responsible and socially grounded technological development.

0 citationsRead paper

Assistive Decision-Making for Right of Way Navigation at Uncontrolled Intersections

Sep 22, 2025

To address the high accident risk at unsignalized intersections—stemming from ambiguous right-of-way, occluded visibility, and unpredictable driver behavior—this paper proposes the first POMDP-based decision-support framework for human-driven vehicles. Methodologically, we develop a high-fidelity simulation platform incorporating stochastic traffic flow, pedestrian dynamics, visual occlusions, and adversarial scenarios, and systematically evaluate three probabilistic planners—QMDP, POMCP, and DESPOT—against a deterministic finite-state machine baseline. Our key contribution is the novel application of POMDPs to right-of-way assistance for human drivers, revealing the critical role of explicit uncertainty modeling in safety-critical decision-making. Experimental results demonstrate that probabilistic planners achieve up to 97.5% collision-free intersection traversal; among them, POMCP attains the highest safety performance, while DESPOT achieves the best trade-off between computational efficiency and real-time feasibility.

0 citationsRead paper