NYU / Columbia / Princeton

Artificial intelligence for science (AI4S) is extending from interpretation of existing data to the design and execution of experiments. Space offers distinctive scientific opportunities while restricting physical support. Microgravity and extraterrestrial environments expose phenomena that terrestrial experiments cannot fully reproduce; finite samples, constrained resources, and limited opportunities for intervention complicate sustained investigation. This critical review examines orbital laboratories and planetary in situ science across biology, materials, fluid and thermal physics, and adaptive exploration. We distinguish learning from space-derived data, selecting scientific actions, executing them, and preserving experimental validity. Terrestrial self-driving laboratories provide complementary evidence on optimization, dedicated automation, robotic manipulation, and the support needed for unattended operation. The synthesis connects scientific gains already demonstrated by bounded AI systems to gaps in observation, executable actions, and continuity of valid evidence. Space robots can expand the attainable scope of autonomous science when necessary interventions exceed validated instrument capabilities and timely human support is unavailable. Their value must be established against dedicated mechanisms, redundancy, and human-supported operation. We develop a framework connecting sample history, scientific decisions, physical execution, and claim-specific validity. A worked orbital design, a planetary extension, and matched evaluation schemes separate the contributions of sensing, decision policies, and physical actions. The resulting agenda measures autonomy through valid scientific return, task coverage, intervention costs, and evidence integrity across complete campaigns.

Scaling supervised fine-tuning for working agents requires not only broad coverage of professional tasks, but also expert demonstrations that faithfully carry out professional workflows to completion. However, missing or inconsistent task information can disrupt expert demonstrations, weakening the resulting supervision. We introduce WorkFactory, a scalable task-construction framework that turns professional intents into executable task environments through hybrid intent sourcing, semantic workflow planning, and execution-based repair. Starting from these intents, WorkFactory constructs a Semantic Workflow Graph that explicitly represents information requirements, dependencies, and locations within the task environment. The graph guides the instantiation of task environments, while an adversarial execution-refinement loop exposes missing or inconsistent information and steers targeted repairs that preserve the original intent. Using WorkFactory, we construct 112,686 tasks spanning 612 occupations and train the WorkPilot family at 4B, 27B, and 35B-A3B scales on the resulting teacher trajectories. All three models improve over their respective base models across four professional-work benchmarks, with WorkPilot-35B-A3B rising from 879 to 1565 Elo on GDPval-AA v2.1. Scaling experiments with up to 110k tasks further show that larger models benefit more from additional task data but exhibit diminishing returns at smaller data scales, while WorkPilot's long-horizon execution evolves from verbose trajectories toward more efficient task completion. Controlled component studies quantify the negative effects of task-environment defects on downstream performance and show how semantic workflow planning and execution-based repair yield downstream gains through improved task-environment quality. The subset of data are available at https://github.com/IQuestLab/WorkFactory.

Long-horizon LLM agents interact with their environments through complete assistant turns while generating each action token by token. We observe that critic values vary primarily across turns, motivating optimization anchored at turns with finer credit allocation among tokens. We propose Multiscale Policy Optimization (MPO), combining Multiscale Advantage Estimation (MAE) and turn-level importance sampling to align credit assignment and policy updates at the turn scale. Extensive experiments across search, tool-use, and automation tasks with 4B and 30B-A3B models demonstrate consistent gains over strong RL baselines.

Capturing a loose object inside a spacecraft requires a robot to approach a target that can translate and rotate, and whose motion changes when contact is made. These conditions motivate evaluation beyond manipulation of supported objects. We introduce AstroBench, an open benchmark for free-floating capture and space robotic manipulation with a fixed-base arm. The task suite combines four initial-motion tracks with sample-storage operations based on MELFI hardware and cap and captive-screw servicing based on the Robotic Refueling Mission. The two workcells provide nine task definitions and 29 task-target combinations. A metadata-filtered Objaverse inventory supplies 8,414 object instances for asset expansion, alongside five cabin visual contexts and versioned task instructions. We report two acquisition-focused experiments using existing checkpoints. A frozen $π_{0.5}$ DROID joint-position policy captures 2 of 200 static targets and 4 of 200 moving targets across twenty object identities. A cube-specific ACT policy captures 8 of 10 static targets and 6 of 10 moving targets in a separate experiment. Each experiment retains its recorded observation, action and capture configuration. The results document difficult direct transfer for the tested generalist checkpoint and successful task-specific capture behavior, while identifying concrete settings for studying motion-aware perception and contact-sensitive control. Code, task definitions and editable workcell assets are available at https://github.com/AstroUpLab/AstroBench.

We present IQuest-Q1, a sparse Mixture-of-Experts (MoE) agentic foundation model with 320B total parameters (15B activated). IQuest-Q1 is developed through a three-stage process: (1) stable and efficient pre-training, combining Stable SplitNorm, Muon, and muP with a two-stage curriculum whose second stage increases the coding-data mixture, preserving a wide stability margin between safe and optimal learning rates; (2) fine-grained mid-training, expanding the data mixture across domains and formats and progressively extending the context; and (3) multi-scaffold with multi-version post-training, combining SFT, domain-specialized RL, and MOPD across various scaffolds to integrate complementary training signals and strengthen agentic coding capabilities. Furthermore, model merging combines different experts and checkpoints from different stages while mitigating conflicts among their parameter updates. The resulting IQuest-Q1 demonstrates capabilities in both coding and general tasks. We release IQuest-Q1 as an open-source model to support community research and development.

Most contemporary artificial intelligence (AI) systems are released with frozen parameters and improve through external updates, whereas an ideal system would keep improving from its own experience, a capacity we call recursive self-improvement (RSI). Yet studying RSI requires an environment that is continuously changing, objectively verifiable, and naturally grounded, posing a fundamental challenge for agent evolution. Here we introduce EvoForesight, a sustainable evolution agent system grounded in future event prediction, in which agents continuously interact with a dynamically evolving environment and improve by internalizing and externalizing the experience gained through these interactions. A living environment synthesizes verifiable tasks, and a recursive-evolution engine turns the reasoning trajectory behind each prediction back into capability: decoupling search from prediction adapts parameter-level evolution to the contamination-prone live environment, while distilling reusable skills lets the agent evolve from the limited live trajectories such an environment affords. On the external FutureX leaderboard, which evaluates predictions against real-world outcomes and is updated weekly, our self-improved forecaster ranked first across all systems over three consecutive weeks, achieving up to a 195% improvement in rating. On our synthesized evaluation, it ranked among the top three across four base LLMs, outperforming the strongest baseline by up to 10.7 Bradley--Terry points. More broadly, these results suggest that a system built for self-improvement is best understood as an agent that co-evolves with a genuinely living, open-ended environment, pointing to a shift in AI from training static models on fixed data toward cultivating agents that grow stronger from their own experience, with such an open world as the indispensable basis of that growth.

Language models are being explored as the interface through which space robots interpret operator commands and select among verified skills. Running them onboard calls for low-bit compression, and flight use adds requirements that generic post-training quantization leaves implicit: a fixed memory budget and storage format, a static artifact under configuration control, and behavior close to the full-precision model validated on the ground. We treat compression as a ground-segment step and study QSelect, which composes one static low-bit model from compatible AWQ, GPTQ, and AutoRound checkpoints by assigning each linear module to one quantizer family and scoring every substitution at the final next-token distribution. The default fidelity mode needs no labels or tuned thresholds and accepts a substitution only when the forward KL divergence from the BF16 reference decreases; an optional target-adapted mode adds answer likelihood on a small labeled set. On DeepSeek-R1-Distill-Qwen-7B, Llama-3.1-8B, and Llama-3.2-3B at 3 and 4 bits, composed models reach lower mean held-out perplexity than every single-family model in all six settings, and the fidelity mode lowers perplexity and KL further. In a matched scaffold, final-output scoring gives lower held-out KL than representation-similarity scoring in all 12 paired runs. With labels from training or development splits of the evaluated task families, the target-adapted mode raises 3-bit pooled accuracy on the two Llama models by 4.3 and 4.5 points over the best single-family model. At 3 bits, unquantized embedding and output layers make up 43-46% of the packed artifact. All evaluations use general-domain benchmarks on a ground GPU; space-domain tasks and flight hardware are left to future work.

Multi-quadrotor passage planning requires coordinated spatial routes and passage times under vehicle dynamics and actuation constraints. Optimizing these decisions jointly couples temporal allocation, nonlinear motion, and inter-vehicle interactions, creating a computational challenge for local trajectory generation. We propose a learning-optimization framework that predicts spatial waypoints and ordered passage events for independently solved quadrotor planners. A common physical time grid aligns the vehicles, with intermediate events fixed by the networks and terminal progress optimized locally to permit different completion metrics. Differentiable sequential convex programming incorporates rigid-body dynamics and rotor-thrust bounds, allowing collision and completion gradients to train both networks through the generated trajectories. In a two-quadrotor evaluation across 24 layouts, the adapted policy passes physical feasibility and sampled separation checks in all five held-out layouts. Their mean completion metric decreases by 1.52% relative to the inherited policy. The median forward time is 21.01 ms on the held-out exchange layout with matching numerical trajectory references. These results demonstrate that learned spatial and event parameters can coordinate independent rigid-body planners while preserving local vehicle constraints in the evaluated scenarios.

Torque-limited spacecraft attitude capture requires adapting the maneuver speed to uncertain braking authority. Faster settings can reduce average tracking error while worsening individual maneuvers, making selective modification of an existing controller a baseline-relative learning problem. A paired-outcome method combines three mechanisms: physically interpretable continuation policies with different speed envelopes, separate prediction of task failure and successful-pair tracking gains, and context-dependent rejection that executes the unchanged baseline. Complete paired simulations provide labels under common plant parameters and measurement seeds; training, validation, and calibration are separated by context. Basilisk experiments use 108 development contexts, 324 labeling rollouts, and 624 evaluation episodes across 48 test contexts, including disclosed post-freeze diagnostics. The selective controller reduces aggregate integrated attitude error by 4.67% relative to the branch baseline, with 17 improvements, 31 exact ties, and no observed regressions. Always using the fast alternative achieves a 4.81% reduction but worsens 11 cases. Gains occur in principal-axis, near-axis, braking-boundary, and selected noise/drift maneuvers; general coupled cases retain the baseline. Ablations show no additional aggregate benefit from branch commitment, active querying, or score subtraction. The evidence supports a restricted empirical trade-off between average improvement and individual regression, without establishing a closed-loop safety guarantee.

Space robots must interpret tasks and act despite unfamiliar dynamics, difficult sensing conditions, and limited opportunities for human intervention. Foundation models offer reusable perceptual and behavioral knowledge, but their value depends on how that knowledge reaches physical execution. This critical review follows space-specific requirements through perception, prediction, planning, and action to deployment and validation, emphasizing orbital manipulation and proximity operations. Terrestrial comparisons establish benefits of particular pretraining stages and model interfaces; their transfer to space dynamics and constrained onboard operation remains insufficiently tested. Event cameras offer timely motion observations across strong illumination contrasts, while spiking neural networks provide possible routes to sparse local computation. We examine how sensor degradation, state estimation and computing choices affect feedback and recovery. Higher action throughput need not yield fresher feedback; lower processor power can also coexist with higher energy per inference. We organize validation as an iterative sim-to-ground-to-space pathway, connecting physical tests with telemetry-based diagnosis and closed-loop revalidation. A satellite-capture example links observations to approach, contact and recovery, while a corresponding protocol separates generalization from adaptation with limited target-domain interaction. These comparisons identify what models already contribute and the physical evidence needed to support autonomous space operation.

Recent advances in large language models (LLMs) have enabled agents to tackle long-horizon tasks across diverse environments. To further improve agent performance, existing language world models typically predict environment observations, yet reconstructing high-entropy, execution-dependent tool responses offers limited value when real feedback is available. Meanwhile, agents suffer from task-state contamination, where unsupported assumptions and outdated plans persist in history and distort subsequent decisions. We propose the Agent-Editing World Model (AEWM), which models how reasoning and actions shape future task progress rather than simulating tool responses. AEWM combines Action Judge to distinguish Critical, Exploratory, and Noisy decisions with State Revision to edit noisy reasoning--action continuations from the same observed history. EditAct integrates these capabilities with real execution, directly changing the state underlying subsequent decisions rather than merely providing critiques. We train AEWM across Search, Terminal, and Software Engineering through mid-training and supervised fine-tuning. AEWM achieves 70.5% macro-F1 on our Action Judge benchmark, exceeding the strongest frontier baseline by 10.6 points. Across six benchmarks and three agent backbones, EditAct improves average scores by 3.2--6.7 points over the strongest baseline. Furthermore, rejection sampling fine-tuning on verified EditAct trajectories, termed AEWM-RFT, improves over Self-RFT by 2.2--2.6 points across three domains without online AEWM guidance.

Planning with world models enables test-time adaptation for embodied agents, offering superior generalization over direct policy learning. However, long-horizon tasks remain hindered by cumulative errors and search complexity. Existing hierarchical methods often use rigid macro-actions or latent skills, leading to brittle representations and limited fine-grained control. We introduce ID-WM (Implicit Dynamics World Models), a hierarchical world model that decouples state-space reachability from local control. Our framework utilizes a high-level model to sample feasible transitions across the state manifold and a low-level world model as an inverse dynamics solver to ground these transitions into precise actions. This allows the planner to identify reachable subgoals without being constrained by low-level execution details. By prioritizing local physical feasibility over global behavioral patterns, ID-WM synthesizes novel long-horizon plans from short segments, transcending the specific behavioral modes seen during training. We demonstrate its effectiveness across three navigation and manipulation suites, achieving a 38.5% improvement over state-of-the-art policy and diffusion-based planners. Crucially, by leveraging this hierarchical structure to mitigate compounding errors, our approach yields a 135% performance gain over single-level world model MPC, particularly in complex, unseen configurations for long-horizon tasks. Videos can be viewed at https://id-wm.github.io

Whether models trained on different modalities converge to the same representation of the world is central to AI. The Platonic Representation Hypothesis (PRH) conjectures that they do as they scale, but without a proof it offers neither a guarantee nor a recipe. We turn the hypothesis into a theorem. First, we prove a Platonic Representation Theorem: Joint-Embedding Predictive Architectures (JEPAs) trained separately on any number of modalities learn the same representation up to one rotation whenever their positive pairs are two observations of one scene--good views--and their misalignment is bounded otherwise. Second, we show that next-token prediction is a JEPA with bad views, parts of a single document, which explains why language models align with vision only through scale. Third, we validate the theory from 145k captions to the texts of 940M web images, and uncover a scaling law of alignment that predicts unseen runs within seed noise. Good views need 42 times fewer training samples than next-token prediction on the same corpus to reach the same alignment, and a 109M-parameter text JEPA aligns with DINOv2 and CLIP's image encoders better than every language model we evaluated, up to 72B parameters, and than CLIP's text encoder: next-token prediction is a remarkably inefficient route to platonic representations. As a showcase, one rotation turns separately trained encoders into an image captioner (CIDEr .838 on COCO). Even that rotation needs no pair: on the latest pair-free benchmark, our text JEPA beats every zero-shot accuracy reported with language models up to 8B parameters, e.g., 64% against at most 56% top-5 on ImageNet-100. Positive views, and not scale alone, are thus the lever to build platonic representations.

Joint-embedding predictive architectures (JEPAs) learn world models directly in latent space, but unconstrained autoregressive predictors may contract the latent distribution over long rollouts. We introduce Symplectic Hamiltonian JEPA (LeHami), which replaces the standard predictor with an action-conditioned Hamiltonian flow over a structured latent state (q, p) integrated using an exactly symplectic leapfrog step. This construction preserves phase-space volume at every rollout step and rules out uniform contraction, asymptotically stable fixed points, and merging of distinct latent states, independently of the learned parameters or action sequence. We obtain the momentum-like component from temporal derivatives of the learned position-like representation, providing an explicit second-order state for prediction and control. On conservative and actuated visual pendulum environments, LeHami substantially improves long-horizon velocity prediction over an unconstrained JEPA world model and empirically maintains constant latent volume where the baseline contracts. The learned flow also supports accurate integration at unseen sub-frame step sizes and numerical time reversal. These results show how geometric structure can provide inference-time guarantees that are complementary to representation regularization in latent world models.

Robotic systems often exhibit unstable modes, along which small perturbations and disturbances can cause unbounded growth unless corrected through feedback. Controlling such systems from high-dimensional visual observations requires representations that preserve these modes. Joint-embedding predictive architectures (JEPAs) provide a natural framework for learning such representations and their dynamics from visual data. However, we demonstrate that next step prediction combined with anti-collapse regularization does not guarantee that controllable unstable modes are preserved: the training loss can be minimized while these modes are collapsed, making stabilization from the learned representation impossible. To address this, we augment world-model training with an action reconstruction objective (i.e., an inverse dynamics loss) that encourages control-aware representations, namely, visual representations that preserve crucial features for control. We prove that exact action reconstruction makes the encoder injective on the finite-horizon reachable subspace. Thus, the encoder cannot discard any state direction reachable by an action sequence within $H$ steps. Moreover, we show that, as $H$ grows, the dominant eigenspace of the finite-horizon controllability Gramian converges to the controllable unstable subspace. We establish our theoretical results for linear systems and demonstrate empirically that our findings extend to nonlinear visual control tasks (CartPole, Walker2D, and PointMaze), highlighting the benefits of control-aware representation learning.

Long-horizon planning with latent world models requires reasoning across timescales and levels of abstraction. Existing task-agnostic JEPA world models predict and plan in a single latent space, often at a single timescale. We introduce H-JEPA, an end-to-end recipe for training a hierarchy of action-conditioned JEPAs in which each level predicts farther ahead in its own learned latent space. Planning proceeds top-down: the top level optimizes progress toward the goal, and each level's predictions become subgoals for the planner below it. When factors in the data evolve at separated timescales, higher levels discard fast, unpredictable detail and retain slower task-relevant state. Across four simulated navigation and manipulation environments, hierarchical planning improves over a flat JEPA; on Visual AntMaze, a three-level hierarchy raises success from 18% to 73% using less planner compute. Ablations attribute these gains to both temporal decomposition and higher-level goal representations. With inverse-dynamics supervision, the approach extends to diverse real-robot videos from DROID, where hierarchy improves offline planning fidelity at lower planner compute.

Modern transformers pair impressive capabilities with substantial memory and compute demands. Low-rank weight factorization reduces both while keeping the matrices dense, and thus efficient on standard hardware. Existing methods, however, choose the subspace to remove from each weight matrix with local closed-form criteria: activation energy, layer-wise reconstruction error, or a quadratic approximation of the loss. These criteria ignore how errors propagate through the network, so at high compression the errors compound with depth and performance collapses. We introduce Learnable Subspace Projections (LSP), which instead learns the subspaces to discard end-to-end. Each linear layer, or tied group of layers that read the same activations, is assigned an orthogonal projector. All projectors are optimized jointly against a global objective--the KL divergence to the dense model's output distribution or the model's original training loss--while the pretrained weights remain frozen. Projectors are initialized from a whitened SVD truncation, and ranks are allocated by the output KL each projector induces per parameter saved. After training, the projectors merge into standard low-rank factors, with each tied group sharing one factor. In attention, this also lets the model cache one narrow latent in place of full keys and values. Across LLMs (OPT-125M/1.3B, Qwen3-4B, Llama-2-7B) and ViT-B/16, LSP outperforms baselines, and its advantage widens as compression increases. At -70% compression, LSP brings Llama-2-7B to 10.9 WikiText-2 perplexity and 42.2% mean zero-shot accuracy, versus 13.3 and 36.0% for the strongest baseline. The factorized model decodes up to 1.6x faster than the dense model at small batch sizes, and aching the shared latent shrinks the combined memory of weights and KV cache by 13.5x at a 128k-token context, versus at most 6.5x for untied baseline factorizations.

Recovering complete physical fields from sparse observations is challenging because the measurements may not uniquely determine the underlying state. Diffusion-based PDE solvers address this problem through iterative sampling whereas neural operators provide deterministic one-pass predictions. We propose SCOPE (Sparse-Context Observability-aware Predictive Embeddings) to recover complete PDE fields from sparse observations by coupling full-field latent prediction with physical reconstruction. A shared decoder reconstructs fields from both predicted and complete-view representations so that representation learning is guided by both physical recovery and latent matching. We derive a quadratic risk decomposition at fixed teacher-decoder pairs showing why optimal latent prediction need not yield optimal field reconstruction. We also establish sufficient conditions for decoder improvements on complete inputs to transfer to recovery from partial observations. Experiments across five PDE settings show that SCOPE outperforms mask-aware neural operators on all ten forward and inverse tasks and achieves lower errors than those reported for diffusion-based solvers including DiffusionPDE and FunDPS. Decoder-only adaptation further improves recovery without retraining the backbone while retaining deterministic single-pass inference.

Physical-field reconstruction and forecasting depend on both measurement density and spatial layout, yet evaluation under a single observation pattern does not characterize performance when that pattern changes. We introduce PDE-OBS, an integrated benchmarking platform spanning numerical data generation, model training, and inference and evaluation under varying observation conditions. It combines 560,000 fields and trajectories from seven partial differential equation families with configurable observation operators and seven adapted baseline methods for stationary reconstruction and short-horizon forecasting. Separating observation construction from physical records allows users to specify parameterized patterns and deterministic mixtures for training and testing while preserving prediction targets and data splits. The evaluation protocol uses references trained for each test pattern to compare models on identical test observations and targets, alongside equal-count groups for spatial-layout comparisons. On a 14,000-record subset, we evaluate 441 trained models under nine test patterns, yielding 3,969 evaluations. Mean cross-pattern error exceeds mean matched-pattern error in all 49 PDE-method pairs, and this finding persists in a configuration-matched subset of 117 models. Denser test observations do not consistently reduce error for a fixed model. Mixed-pattern training on five completed pairs reduces large single-pattern transfer errors, although destination-trained references usually remain more accurate. Together, the benchmark and findings support systematic evaluation of observation-pattern sensitivity and provide a reusable workflow for developing methods under changing measurement conditions. Code: https://github.com/ru1ch3n/PDE-OBS.

Autonomous driving requires world models that can understand the physical world, reason and plan, and operate safely. In this paper, we first systematically evaluate existing action-conditioned joint-embedding predictive architecture (JEPA) world models, including LeWM, DINO-WM, and JEPA-WM for end-to-end autonomous driving (E2EAD). To isolate world-model quality from policy learning, we employ a goal-conditioned zero-shot planning setting that evaluates these models using ground-truth future observations as goals, without training any driving policy. We find that existing JEPA-based world models are either accurate for driving but computationally expensive, or computationally efficient but insufficient for planning. To address this trade-off, we propose AD-E2E-JEPA, which introduces a SIGReg-regularized learnable projector applied to projected patch embeddings. The projector reduces the number of planning patches by $16\times$ and the embedding dimension by $4\times$, achieving a $100\times$ inference speedup while retaining planning performance, with a 0.8-second runtime for an 8-frame rollout over 256 candidate trajectories. Without training any driving policy, the world model itself reaches the goals located 20 meters away on average within the displacement of respectively 4.0/2.8 meters, using world-model rollouts over trajectory vocabularies of respectively 256/8,192 candidates. On the NAVSIMv2 benchmark, it achieves 67.3/72.9 EPDMS with multiplicative safety metrics and 84.1/86.5 EPDMS$^{\dagger}$ without them in goal-conditioned zero-shot planning. Experiments further show that the self-supervised pretrained projector improves downstream imitation learning performance from 80.2 to 85.4 EPDMS. The source code is available at https://github.com/HaoranZhuExplorer/AD-E2E-JEPA

Contact-rich manipulation benefits from tactile feedback that captures both detailed contact geometry and interaction forces. When scooping grains, contact patterns help infer spoon orientation, while resistance transmitted through the grasp guides motion. To provide these complementary sensing capabilities, we present PanoTac, a low-cost, compliant tactile fingertip designed for dense three-axis tactile sensing. Each fingertip provides a $12 \times 32 \times 3$ tactile representation, with one normal ($z$) and two shear ($x$, $y$) response components at each taxel. The sensor operates at 60 Hz and costs less than $50 per fingertip. Sensor characterization quantifies response repeatability and consistency across sensor instances and contact locations. We integrate PanoTac into a robotic system with a visuo-tactile pipeline. Across test-tube preparation, grain scooping, and salt pouring, the resulting policies outperform vision-only baselines by an average of 46.7 percentage points. Ablations show that removing either spatial contact information or shear feedback reduces overall performance across the task suite. Our project page is available at https://panotac.github.io.

Planning with world models enables test-time adaptation for embodied agents, offering superior generalization over direct policy learning. However, long-horizon tasks remain hindered by cumulative errors and search complexity. Existing hierarchical methods often use rigid macro-actions or latent skills, leading to brittle representations and limited fine-grained control. We introduce ID-WM (Implicit Dynamics World Models), a hierarchical world model that decouples state-space reachability from local control. Our framework utilizes a high-level model to sample feasible transitions across the state manifold and a low-level world model as an inverse dynamics solver to ground these transitions into precise actions. This allows the planner to identify reachable subgoals without being constrained by low-level execution details. By prioritizing local physical feasibility over global behavioral patterns, ID-WM synthesizes novel long-horizon plans from short segments, transcending the specific behavioral modes seen during training. We demonstrate its effectiveness across three navigation and manipulation suites, achieving a 38.5% improvement over state-of-the-art policy and diffusion-based planners. Crucially, by leveraging this hierarchical structure to mitigate compounding errors, our approach yields a 135% performance gain over single-level world model MPC, particularly in complex, unseen configurations for long-horizon tasks. Videos can be viewed at https://id-wm.github.io

Current recipes for robot learning require massive amounts of robot demonstrations to obtain robust and adaptive policies. In this work, we introduce Steertuning, a system that enables robots to learn new tasks in new environments from 5 human demonstrations in under 5 Steertuning consists of two parts. (1) We pretrain a 400M-parameter diffusion policy on our egocentric human dataset, which consists of 100 hours of densely annotated random interactions. This pretraining makes Steertuning robust to background differences and enables spatial and object generalization when deployed on new tasks. (2) Given 5 demonstrations of a new task, we finetune our model for only 2 minutes (2000 training updates) and steer the policy's actions during reverse sampling using the demonstration whose observation is closest to that of the deployment environment. Together with a carefully designed data processing pipeline for embodiment transfer, this enables Steertuning to achieve a 71% success rate across 12 tasks in 12 new environments, ranging from lab settings to kitchens in new homes. The entire process, from data collection for a new task to robot policy deployment, takes less than 5 minutes. Our full dataset, model weights, codebase, and robot videos will be made available at our website: https://steertuning.github.io.