Publications

Large language models (LLMs) often exhibit sycophancy, optimizing for agreement over productive challenge, which severely limits their utility in domains like professional skills training, where growth requires pushback. We introduce, ConvoDojo, a novel conversational AI platform for practicing difficult workplace conversations, engineered not merely as a commercial training application but also as a flexible, instrumented research platform for evaluating conversational AI strategies. ConvoDojo repurposes LLMs as structured sparring partners to support skill development in difficult workplace conversations (e.g., performance feedback, conflict resolution), addressing the reported managerial tendency to avoid them. This paper showcases the platform and presents an evaluation of how key conversational user interface (CUI) design elements, namely, the addition of structured feedback and upfront instructional scaffolding, impact managers’ learning. Results show that ConvoDojo is highly engaging and promotes user reflection. We demonstrate how theory-informed dialogue and adaptive pushback can transform an LLM into an effective, measurable tool for complex communication skills development.

Proximal methods such as the Alternating Direction Method of Multipliers (ADMM) are effective at solving constrained quadratic programs (QPs). To tackle infeasible QPs, slack variables are often introduced to ensure feasibility, which changes the structure of the problem, increases its size, and slows down numerical resolution. In this letter, we propose a simple ADMM scheme to tackle QPs with slack variables without increasing the size of the original problem. The only modification is a slightly different projection in the z-update, while the rest of the algorithm remains standard. We prove that the method is equivalent to applying ADMM to the QP with additional slack variables, even though slack variables are not added. Numerical experiments show speedups of the approach.

Curated datasets are essential for training and evaluating AI approaches, but are often lacking in domains where language and physical action are deeply intertwined. In particular, few datasets capture how people acquire embodied skills through verbal instruction over time. To address this gap, we introduce SIMCOACHCORPUS: a unique dataset of race car simulator driving that allows for the investigation of rich interactive phenomena during guided and unguided motor skill acquisition. In this dataset, 29 humans were asked to drive in a simulator around a race track for approximately ninety minutes. Fifteen participants were given personalized one-on-one instruction from a professional performance driving coach, and 14 participants drove without coaching. SIMCOACHCORPUS includes embodied features such as vehicle state and inputs, map (track boundaries and raceline), and cone landmarks. These are synchronized with concurrent verbal coaching from a professional coach and additional feedback at the end of each lap. We further provide annotations of coaching categories for each concurrent feedback utterance, ratings on students' compliance with coaching advice, and self-reported cognitive load and emotional state of participants (gathered from surveys during the study). The dataset includes over 20,000 concurrent feedback utterances, over 400 terminal feedback utterances, and over 40 hours of vehicle driving data. Our naturalistic dataset can be used for investigating motor learning dynamics, exploring linguistic phenomena, and training computational models of teaching. We demonstrate applications of this dataset for in-context learning, imitation learning, and topic modeling.

We propose a fast and correspondence-free local point cloud registration method that leverages geometric surface structure and reproducing kernel Hilbert space (RKHS) embeddings. The method represents point clouds as continuous functions with point-wise anisotropic kernels that encode local geometry. This formulation improves alignment along surface normals while relaxing alignment along tangential directions. To solve the resulting registration problem, we propose a second-order on-manifold optimization scheme with approximate Riemannian Hessians, achieving a speedup of up to 10x over the first-order solvers used in prior correspondence-free RKHS-based methods. We demonstrate improved frame-to-frame LiDAR and RGB-D tracking accuracy across diverse indoor and outdoor datasets. On a LiDAR tracking registration task in the driving domain, we achieve a reduction of > 55% in both translational and rotational drift in challenging feature-sparse environments. On object registration benchmarks, we show improved robustness over ICP-based methods and further gains when refining global initialization, particularly under moderate misalignment.

Gaze behavior is widely treated as a trainable component of high-performance driving, yet its causal role in performance remains unclear. We tested whether enforcing expert gaze patterns improves circuit driving performance in a simulator study with 60 participants assigned to free gaze, skilled-gaze guidance, or novice-gaze guidance. Gaze guidance replayed naturalistic gaze trajectories from actual skilled and novice drivers during training, followed by an unguided retention session. Despite clear compliance with gaze instructions during training, gaze guidance had no significant effect on lap time, steering or pedal smoothness, or lateral deviation from an optimal racing line, aside from a minimal per-corner difference. Performance improvements were attributable to practice and persisted independently of gaze condition. These findings suggest that gaze guidance alone is insufficient to improve high-performance driving, underscoring the need for training approaches that integrate it with additional instruction or feedback.

We propose a new approach for solving planning problems with a hierarchical structure, fusing reinforcement learning and MPC planning. Our formulation tightly and elegantly couples the two planning paradigms. It leverages reinforcement learning actions to inform the MPPI sampler, and adaptively aggregates MPPI samples to inform the value estimation. The resulting adaptive process leverages further MPPI exploration where value estimates are uncertain, and improves training robustness and the overall resulting policies. This results in a robust planning approach that can handle complex planning problems and easily adapts to different applications, as demonstrated over several domains, including race driving, modified Acrobot, and Lunar Lander with added obstacles. Our results in these domains show better data efficiency and overall performance in terms of both rewards and task success, with up to a 72% increase in success rate compared to existing approaches, as well as accelerated convergence (x2.1) compared to non-adaptive sampling.

Effective coaching requires instruction to be tailored to each student’s skill level. While prior work has shown the importance of adaptive feedback, there is little quantitative evidence showing how coaching language itself changes as a function of skill level - particularly in real-time embodied domains. To address this gap, we collected a naturalistic dataset of an expert coach providing verbal instructions to students of varying skill in a high-performance driving simulation. Using linguistic and interaction measures, we show that, as skill improves, coaching language shifts from concrete, instructional feedback, towards evaluative and affective language. At the same time, while overall feedback amount decreases, language diversity increases, indicating a move toward denser, more expressive communication. Together, these findings provide us with a better understanding of how a human coach modulates language as a function of learner skill and provide empirical grounding for the design of adaptive and personalized AI coaching systems.

Generative world models offer a compelling foundation for augmented-reality (AR) applications: by predicting future image sequences that incorporate deliberate visual edits, they enable temporally coherent, augmented future frames that can be computed ahead of time and cached, avoiding per-frame rendering from scratch in real time. In this work, we present SEGAR, a preliminary framework that combines a diffusion-based world model with a selective correction stage to support this vision. The world model generates augmented future frames with region-specific edits while preserving others, and the correction stage subsequently aligns safety-critical regions with real-world observations while preserving intended augmentations elsewhere. We demonstrate this pipeline in driving scenarios as a representative setting where semantic region structure is well defined and real-world feedback is readily available. We view this as an early step toward generative world models as practical AR infrastructure, where future frames can be generated, cached, and selectively corrected on demand.

Short-range order (SRO) in the cation-disordered state is a controlling factor influencing the probability of finding tetrahedron clusters in disordered rocksalt (DRX) cathode materials. However, the prevalent  probability below the random limit across reported DRX compositions has not been systematically investigated, active strategies to surpass the random limit of  probability are lacking, and the fundamental ordering behavior on the face-centered cubic (FCC) lattice remains insufficiently explored. This research quantitatively examines pair SRO parameters and  probabilities via exhaustive Monte Carlo mapping across a simplified subset of the parameter space. The results indicate that, in the disordered state, the  probability is governed by the nearest neighbor (NN) pairwise SRO parameter, and that these quantities do not necessarily represent a simple attenuation of their corresponding low-temperature long-range order, particularly for the important cases of Layered and Spinel-like orderings. Strategies are proposed to mitigate or even reverse the lithium and transition metals mixing tendency of NN pair SRO to achieve  probabilities that exceed the random limit. This study advances the fundamental thermodynamic understanding of ordering behaviors, which can be generalized to any FCC system.

Embodied foundation models are increasingly performant in real-world domains such as robotics or autonomous driving. These models are often deployed in interactive or assistive settings, where it is important that these assistive models generalize to new users and new tasks. Diverse interactive data generation offers a promising avenue for providing data-efficient generalization capabilities for interactive embodied foundation models. In this paper, we investigate the generalization capabilities of a multimodal foundation model fine-tuned on diverse interactive assistance data in a synthetic domain. We explore generalization along two axes: a) assistance with unseen categories of user behavior and b) providing guidance in new configurations not encountered during training. We study a broad capability called Open-Set Corrective Assistance, in which the model needs to inspect lengthy user behavior and provide assistance through either corrective actions or language-based feedback. This task remains unsolved in prior work, which typically assumes closed corrective categories or relies on external planners, making it a challenging testbed for evaluating the limits of assistive data. To support this task, we generate synthetic assistive datasets in Overcooked and fine-tune a LLaMA-based model to evaluate generalization to novel tasks and user behaviors. Our approach provides key insights into the nature of assistive datasets required to enable open-set assistive intelligence. In particular, we show that performant models benefit from datasets that cover different aspects of assistance, including multimodal grounding, defect inference, and exposure to diverse scenarios.