Publications

Frontier models are now used for browsing, self-driving, assembly work and tabletop manipulation tasks. Their performance on these domains and deployment decisions rely on the benchmark scores labs report. However, a benchmark mean hides how much a model's

Satellite constellations require orbital transfers that are both fuel efficient and collision avoidant. Yet, the computational cost of optimization methods traditionally used to plan their trajectories scales poorly with both the number of satellites as well as the number of

We evaluated 3 highly capable Vision-Language Models (VLMs) on 50 2D mazes. Across the 3 models, only 6 total mazes were solved, and 45 of the 50 mazes were not solved by any model.

The size of the mazes

Structural damage in modular spacecraft can disrupt mechanical and communication connectivity, reducing system capability. Existing approaches rely on redundancy or preplanned reconfiguration and do not enable autonomous repair under local information and physical constraints. We model the spacecraft as a

GUI grounding models report over 85% accuracy on standard benchmarks, yet drop 27-56 percentage points when instructions require spatial reasoning rather than direct element naming. Current benchmarks miss this because they evaluate each screenshot once with a single fixed

A key challenge in relative orbital motion is the design of nonlinear controllers that provide formal stability guarantees for the nonlinear dynamics while systematically incorporating operational constraints such as actuator saturation and sensor field-of-view requirements. In this paper,

Multimodal reasoning and action models hold immense promise as general-purpose agents, yet the current evaluation landscape remains fragmented with domain-specific benchmarks that fail to capture true generalization capabilities. This critical gap prevents us from understanding where these sophisticated

Recent innovations in multimodal action models represent a promising direction for developing general-purpose agentic systems, combining visual understanding, language comprehension, and action generation. We introduce MultiNet - a novel, fully open-source benchmark and surrounding software ecosystem designed to rigorously

Vision-language-action (VLA) models represent an important step toward general-purpose robotic systems by integrating visual perception, language understanding, and action execution. However, systematic evaluation of these models, particularly their zero-shot generalization capabilities in procedurally out-of-distribution

The expanding space economy has created an urgent need for reliable on-orbit transportation systems to support both commercial and scientific missions. This paper explores a cooperative approach to orbital transportation, where a swarm of spacecraft agents, each securely attached

We propose affine concept editing (ACE) as an approach for steering language models' behavior by intervening directly in activations. We begin with an affine decomposition of model activation vectors and show that prior methods for steering model behavior correspond to

Below is the abstract for our recent paper, "Intelligent Digital Agents in the Era of Large Language Models". We’re growing our Research Team and pursuing new projects. If you’re interested in working together, join the conversation on 

Conversational AI agents often require extensive datasets for training that are not publicly released, are limited to social chit-chat or handling a specific domain, and may not be easily extended to accommodate the latest advances in AI technologies. This

Recent advances in recurrent neural network architectures, such as Mamba and RWKV, have enabled RNNs to match or exceed the performance of equal-size transformers in terms of language modeling perplexity and downstream evaluations, suggesting that future systems may be

Purpose

This study aims to report on an automatic deidentification system for labeling and obfuscating personally identifiable information (PII) in student-generated text.

Design/methodology/approach

The authors evaluate the performance of their deidentification system on two data sets of

How do we design an AI system that is intended to act as a communication bridge between two user communities with different mental models and vocabularies? Skillsync is an interactive environment that engages employers (companies) and training providers (colleges) in

As education increasingly takes place in technologically mediated settings, it has become easier to collect student data that would be valuable to researchers. However, much of this data is not available due to concerns surrounding the protection of student privacy.

Machine Teaching (MT) is an interactive process where a human and a machine interact with the goal of training a machine learning model (ML) for a specified task. The human teacher communicates their task expertise and the machine student gathers

Deep neural network based face recognition models have been shown to be vulnerable to adversarial examples. However, many of the past attacks require the adversary to solve an input-dependent optimization problem using gradient descent which makes the attack impractical

We describe a stance towards the generation of explanations in AI agents that is both human-centered and design-based. We collect questions about the working of an AI agent through participatory design by focus groups. We capture an agent'