To serve global users safely and productively, LLMs need culture-specific knowledge that might not be learned during pre-training. How do we find such knowledge that is (1) salient to in-group users, but (2) unknown to LLMs? The most common solutions are single-initiative: either researchers define challenging questions that users passively answer (traditional annotation), or users actively produce data that researchers structure as benchmarks (knowledge extraction). The process would benefit from mixed-initiative collaboration, where users guide the process to meaningfully reflect their cultures, and LLMs steer the process towards more challenging questions that meet the researcher's goals. We propose a mixed-initiative methodology called CultureCartography. Here, an LLM initializes annotation with questions for which it has low-confidence answers, making explicit both its prior knowledge and the gaps therein. This allows a human respondent to fill these gaps and steer the model towards salient topics through direct edits. We implement this methodology as a tool called CultureExplorer. Compared to a baseline where humans answer LLM-proposed questions, we find that CultureExplorer more effectively produces knowledge that leading models like DeepSeek R1 and GPT-4o are missing, even with web search. Fine-tuning on this data boosts the accuracy of Llama-3.1-8B by up to 19.2% on related culture benchmarks.

With an increase in aerial vehicle applications, passive deformable quadrotors are getting significant attention in the research community due to their potential to perform physical interaction tasks. Such quadrotors are capable of undergoing collisions, both planned and unplanned, which are harnessed to induce deformation and retain stability by dissipating collision energies. In this article, we utilize one such passive deforming quadrotor, XPLORER, to complete various contact-rich tasks by exploiting its compliant chassis via various impact-aware planning and control algorithms. At the core of these algorithms is a novel external wrench estimation technique developed specifically for the unique multi-linked structure of XPLORER's chassis. The external wrench information is then employed for designing interaction controllers to obtain three additional flight modes: static-wrench application, disturbance rejection and yielding to the disturbance. These modes are then incorporated into a novel online exploration scheme to enable navigation in unknown flight spaces with only tactile feedback and generate a map of the environment without requiring additional sensors. Experiments show the efficacy of this scheme to generate maps of the previously unexplored flight space with an accuracy of 96.72%. Finally, we develop a novel collision-aware trajectory planner (CATAAN) to generate minimum time maneuvers for waypoint tracking by integrating collision-induced state jumps for both elastic and inelastic cases. We experimentally validate that minimum time trajectories can be obtained with CATAAN leading to a 40.38% reduction of settling time accompanied by improved tracking performance of a root mean squared error in position within 0.5cm as compared to 3cm of conventional methods.

With the rapid commoditization of wearable sensors, detecting human movements from sensor datasets has become increasingly common over a wide range of applications. To detect activities, data scientists iteratively experiment with different classifiers before deciding which model to deploy. Effective reasoning about and comparison of alternative classifiers are crucial in successful model development. This is, however, inherently difficult in developing classifiers for sensor data, where the intricacy of long temporal sequences, high prediction frequency, and imprecise labeling make standard evaluation methods relatively ineffective and even misleading. We introduce Track Xplorer, an interactive visualization system to query, analyze, and compare the predictions of sensor-data classifiers. Track Xplorer enables users to interactively explore and compare the results of different classifiers, and assess their accuracy with respect to the ground-truth labels and video. Through integration with a version control system, Track Xplorer supports tracking of models and their parameters without additional workload on model developers. Track Xplorer also contributes an extensible algebra over track representations to filter, compose, and compare classification outputs, enabling users to reason effectively about classifier performance. We apply Track Xplorer in a collaborative project to develop classifiers to detect movements from multisensor data gathered from Parkinson's disease patients. We demonstrate how Track Xplorer helps identify early on possible systemic data errors, effectively track and compare the results of different classifiers, and reason about and pinpoint the causes of misclassifications.