@inproceedings{hri25.pdf,title={Human Demonstrations Enable Efficient Solutions to Sequential Manifold Planning Problems},author={Crockett, Breanne and Mueller, Carl and Hayes, Bradley},booktitle={Proceedings of the ACM/IEEE International Conference on Human-Robot Interaction (HRI 2025)},location={Melbourne, Australia},month=mar,year={2025},}
2024
Iteratively Adding Latent Human Knowledge within Trajectory Optimization Specifications Improves Learning and Task Outcomes
Christine
Chang, Maria P.
Stull, Breanne
Crockett, Emily
Jensen, Clare
Lohrmann, Mitchell
Hebert, and Bradley
Hayes
IEEE Robotics and Automation Letters (RA-L), Nov 2024
@article{chang2024iteratively,title={Iteratively Adding Latent Human Knowledge within Trajectory Optimization Specifications Improves Learning and Task Outcomes},author={Chang, Christine and Stull, Maria P. and Crockett, Breanne and Jensen, Emily and Lohrmann, Clare and Hebert, Mitchell and Hayes, Bradley},journal={IEEE Robotics and Automation Letters (RA-L)},month=nov,year={2024},}
ORNL Slicer 2 - Open Source Copyright
Alex
Roschli, Michael
Borish, Abigail
Barnes, Charles
Wade, Breanne
Crockett, Liam
White, Cameron
Adkins, USDOE Office
Energy Efficiency, and Renewable
Energy
@misc{slicer2-copyright,title={ORNL Slicer 2 - Open Source Copyright},author={Roschli, Alex and Borish, Michael and Barnes, Abigail and Wade, Charles and Crockett, Breanne and White, Liam and Adkins, Cameron and of Energy Efficiency, USDOE Office and Energy, Renewable},abstractnote={ORNL Slicer 2 is a slicing program for additive manufacturing. It takes a solid body mesh object, typically as .STL file, and converts that into machine readable instructions, called g-code, that a 3D printer can use to build the object. The functionality includes loading and positioning an object, slicing it into layers, fitting toolpaths to the layers, and outputting g-code to construct the object. All of this is contained within a graphical user interface (GUI) that allows the user to define all of the settings specific to their machine and process, then preview the resultant g-code before starting the printing process.},url={https://www.osti.gov/biblio/2352739},doi={10.11578/dc.20240520.1},year={2024},month=may,}
2023
Effective Human-in-the-loop Control Handover via Confidence-Aware Autonomy
Breanne
Crockett*, Kyler
Ruvane*, Matthew B
Luebbers, and Bradley
Hayes
Proceedings of the Workshop on Life-Long Learning with Human Help (L3H2 2023), London, UK, May 2023
Robotics is traditionally divided into two operational paradigms – autonomous control and teleoperation. Both approaches are affected by the inherent strengths and weaknesses of autonomous systems and human operators. Therefore, it is beneficial for many tasks to blend the two operation strategies, incorporating human-in-the-loop supervision with autonomous control. In this work, we explore the question of control handover: when should a robot act autonomously, when should a human supervisor take control, and who should decide this? We first analyze four candidate metrics for estimating confidence in a policy learned via reinforcement learning (count of examples, choice difficulty, Gaussian choice difficulty, and historical upper confidence bound), using those metrics to autonomously trigger handover requests to a human supervisor whenever a robot’s confidence is low. Through a simulation evaluation, we found historical upper confidence bound to be the most correct metric, achieving the highest accuracy on the timing of handover requests. Using this finding, we conducted a human-subjects evaluation, showing that in a human-supervised robotic navigation task, robot-to-human handover triggered autonomously using our method outperformed human-initiated handover, both on robotic task performance and on subjective human measures of workload and usability.
@article{confidence-based-handover,title={Effective Human-in-the-loop Control Handover via Confidence-Aware Autonomy},author={Crockett, Breanne and Ruvane, Kyler and Luebbers, Matthew B and Hayes, Bradley},location={London, UK},journal={Proceedings of the Workshop on Life-Long Learning with Human Help (L3H2 2023)},year={2023},month=may,}
Determining Optimal Print Orientation Using GPU-Accelerated Convex Hull Analysis
Charles
Wade, Breanne
Crockett, Michael
Borish, and Robert
Maccurdy
In Proceedings of the 8th ACM Symposium on Computational Fabrication, New York City, NY, USA, Oct 2023
In fused filament fabrication (FFF), the orientation of a part within the printer volume can dramatically affect print quality and probability of success. An object’s orientation determines how much support structure will be required and the strength of adhesion between the deposited material and the build surface. Selecting a part’s orientation is a non-trivial problem that users of FFF slicing software face routinely. Numerous part orientations need to be considered to find the best according to the results of the slicing process. This paper presents a method to automatically determine an optimal printing orientation for FFF that maximizes build-surface adhesion while minimizing the need for support structure. The algorithm considers the slicing angle and a configurable angle for overhang that requires supporting structure. By employing GPU acceleration and convex hull analysis to limit candidate orientations, the algorithm can run in real time as a preprocessing aid to users slicing parts.
@inproceedings{optimal-print-orientation-convex-hull,author={Wade, Charles and Crockett, Breanne and Borish, Michael and Maccurdy, Robert},title={Determining Optimal Print Orientation Using GPU-Accelerated Convex Hull Analysis},month=oct,year={2023},isbn={9798400703195},publisher={Association for Computing Machinery},address={New York, NY, USA},url={https://doi.org/10.1145/3623263.3623360},doi={10.1145/3623263.3623360},booktitle={Proceedings of the 8th ACM Symposium on Computational Fabrication},articleno={4},numpages={9},keywords={toolpath-planning, optimization, computer-aided design, additive manufacturing, GPU acceleration},location={New York City, NY, USA},series={SCF '23},}
Clustering of Animation View Times in an Interactive Textbook for Material and Energy Balances
Tanner
Hilsabeck, Breanne
Crockett, Amir
Parsaei, Kevin S.
Xu, and Matthew W.
Liberatore
Data science tools can help elucidate trends from clickstreams and other interactions generated by students actively using interactive textbooks. Specifically, data generated when using animations, which are multi-step visuals with text captions, will be presented in this work. Each animation step divides content into appropriate chunks, and so aligns with tenets of cognitive load theory. Both the quantity and timing of students’ clicks record provide large data sets when examining students across hundreds of animations and multiple cohorts. Specifically, an interactive textbook for a chemical engineering course in Material and Energy Balances will be examined and build upon data presented previously. While most of the previous data focused on very high reading completion rates (>99% median) compared to traditional textbooks (20-50%), a deeper examination of how long students take when watching animations will be explored. With over 140 unique animations and tens of thousands of completed views over five cohorts, a spectral clustering algorithm applied to students’ animation view times distinguished several types of animation watching behavior as well as monitor changes in this animation watching behavior over the course of a semester. After examining different numbers of clusters, two or three clusters in each chapter captured the animation usage. These clusters usually correspond to a group of students who watched animations at 1x speed (longer), another group who watched at 2x speed (shorter), and a third group, when present, who watched irregularly, including skipping animations. Overall, more students belonged to the belonged to the cluster with longer view times, with 63% of students aggregated over all cohorts and chapters compared to 35% of students in the cluster with shorter view times. The remaining 2% of students belonged to the irregular cluster, which was present in less than one quarter of the chapters. Many students stayed in the same cluster between chapters, while a smaller fraction switched between the longer and shorter clusters.
@article{interactive-textbook-clustering,title={Clustering of Animation View Times in an Interactive Textbook for Material and Energy Balances},url={https://par.nsf.gov/biblio/10451662},journal={ASEE Annual Conference proceedings},author={Hilsabeck, Tanner and Crockett, Breanne and Parsaei, Amir and Xu, Kevin S. and Liberatore, Matthew W.},year={2023},month=jun,}
2022
Toolpath Planning for Multiple Build Points Using K-Means Clustering
Breanne
Crockett, and Michael
Borish
2022 International Solid Freeform Fabrication Symposium, Jun 2022
Traditional 3D printers deposit material at one build point at a time, often resulting in long print times. To reduce print time, 3D printers could increase throughput with parallel construction at multiple build points. The primary challenge in path planning for parallel construction is dividing an object between the build points. The object should be divided such that the workload is balanced, and the individual build areas are discrete. This work proposes a variation of k-means clustering for object division. The algorithm considers coordinate position and geometric area as an indicator of workload. This method is demonstrated on several test models to compare workload across the number of build points.
@article{multi-build-point-slicer,title={Toolpath Planning for Multiple Build Points Using K-Means Clustering},url={https://hdl.handle.net/2152/117310},doi={http://dx.doi.org/10.26153/tsw/44191},journal={2022 International Solid Freeform Fabrication Symposium},author={Crockett, Breanne and Borish, Michael},year={2022},month=jun,}
Animation Analytics in an Interactive Textbook for Material and Energy Balances
S. J.
Stone, B.
Crockett, Kevin S.
Xu, and M. W.
Liberatore
Interactive textbooks generate big data through student reading participation, including animations, question sets, and auto-graded homework. Animations are multi-step, dynamic visuals with text captions. By dividing new content into smaller chunks of information, student engagement is expected to be high, which aligns with tenets of cognitive load theory. Specifically, students’ clicks are recorded and measure usage, completion, and view time per step and for entire animations. Animation usage data from an interactive textbook for a chemical engineering course in Material and Energy Balances accounts for 60,000 animation views across 140+ unique animations. Data collected across five cohorts between 2016 and 2020 used various metrics to capture animation usage including watch and re-watch rates as well as the length of animation views. Variations in view rate and time were examined across content, parsed by book chapter, and five animation characterizations (Concept, Derivation, Figures and Plots, Physical World, and Spreadsheets). Important findings include: 1) Animation views were at or above 100% for all chapters and cohorts, 2) Median view time varies from 22 s (2-step) to 59 s (6-step) - a reasonable attention span for students and cognitive load, 3) Median view time for animations characterized as Derivation was the longest (40 s) compared to Physical World animations, which resulted in the shortest time (20 s).
@article{textbook-animation-analytics,title={Animation Analytics in an Interactive Textbook for Material and Energy Balances},url={https://par.nsf.gov/biblio/10451700},journal={ASEE Annual Conference proceedings},author={Stone, S. J. and Crockett, B. and Xu, Kevin S. and Liberatore, M. W.},year={2022},month=aug,}
Measuring Students’ Engagement with Digital Interactive Textbooks by Analyzing Clickstream Data
Breanne
Crockett
Proceedings of the AAAI Conference on Artificial Intelligence. AAAI Undergraduate Consortium
, Jun 2022
This paper provides an overview of my contributions to a project to measure and predict student’s mental workload when using digital interactive textbooks. The current work focuses on analysis of clickstream data from the textbook in search of viewing patterns among students. It was found that students typically fit one of three viewing patterns. These patterns can be used in further research to inform creation of new interactive texts for improved student success.
@article{clickstream-interactive-textbooks,title={Measuring Students’ Engagement with Digital Interactive Textbooks by Analyzing Clickstream Data},volume={36},url={https://ojs.aaai.org/index.php/AAAI/article/view/21702},doi={10.1609/aaai.v36i11.21702},journal={Proceedings of the AAAI Conference on Artificial Intelligence},number={11},author={Crockett, Breanne},year={2022},month=jun,pages={13132-13133},}
