Search for: All records

This study examines the content and layout of the proposed broadband consumer disclosure labels mandated by the U.S. Federal Communications Commission (FCC). Our large-scale user study identifies key consumer preferences and comprehension factors through a two-phase survey of 2,500 broadband internet consumers. Findings reveal strong support for broadband labels, but dissatisfaction with the FCC's proposed labels from 2016. Participants generally struggled to use the label for cost computations and plan comparisons. Technical terms confused participants, but providing participants with brief education made the terms usable. Participants desired additional information, including reliability, speed measures for both periods when performance is “normal” and periods when performance is much worse than normal, quality-of-experience ratings, and detailed network management practices. This feedback informed our improved label designs that outperformed the 2016 labels in comprehension and preference. Overall, consumers valued clear pricing and performance details, comprehensive information, and an easy-to-understand format for plan comparison. Requiring broadband service providers to deposit machine-readable plan information in a publicly accessible database would enable third parties to further customize how information is presented to meet these consumer needs. Our work additionally highlights the need for user studies of labels to ensure they meet consumer demands. 

Manipulating volumetric deformable objects in the real world, like plush toys and pizza dough, brings substantial challenges due to infinite shape variations, non-rigid motions, and partial observability. We introduce ACID, an action-conditional visual dynamics model for volumetric deformable objects based on structured implicit neural representations. ACID integrates two new techniques: implicit representations for action-conditional dynamics and geodesics-based contrastive learning. To represent deformable dynamics from partial RGB-D observations, we learn implicit representations of occupancy and flow-based forward dynamics. To accurately identify state change under large non-rigid deformations, we learn a correspondence embedding field through a novel geodesics-based contrastive loss. To evaluate our approach, we develop a simulation framework for manipulating complex deformable shapes in realistic scenes and a benchmark containing over 17,000 action trajectories with six types of plush toys and 78 variants. Our model achieves the best performance in geometry, correspondence, and dynamics predictions over existing approaches. The ACID dynamics models are successfully employed for goal-conditioned deformable manipulation tasks, resulting in a 30% increase in task success rate over the strongest baseline. Furthermore, we apply the simulation-trained ACID model directly to real-world objects and show success in manipulating them into target configurations. https://b0ku1.github.io/acid/