Position: Good Embodied Reward Models Need Bad Behavior Data

This position paper argues that to obtain reliable embodied reward models, the community must invest in ``bad'' robot data: failed, suboptimal, error-prone, and even hazardous behaviors. While reward models are central to any foundation model's lifecycle, today's embodied reward models are trained primarily on successful behaviors. We analyze three state-of-the-art embodied reward models and find that they systematically over-reward behaviors that real human evaluators would penalize, including unsafe interactions, poor execution, and shortcut strategies that only superficially satisfy tasks. We attribute these failures to a key data gap: the scarcity of negative embodied data which is costly to collect and often filtered out or withheld in existing robotics datasets. Furthermore, we show that even modest exposure to real bad behavior data can improve alignment with human preferences and reduce costly false positives. We therefore call on the embodied AI community to curate and release their bad robot data, build synthetic bad data generation engines, develop more decentralized physical evaluation systems, and design benchmarks for fine-grained embodied reward model evaluations.

StressDream: Steering Video World Models for Robust Policy Evaluation and Improvement

Video world models (WMs) have shown promise for policy evaluation and improvement by imagining realistic future observations conditioned on ego-robot actions. While WMs can model distributions over futures, policy evaluation and improvement typically rely on nominal imaginations, which can miss high-impact outcomes of robot actions unless prohibitively many samples are drawn. To enable robust policy evaluation and improvement over WM imaginations, we propose StressDream, which steers imaginations toward high-impact yet plausible outcomes specified at inference time by optimizing the initial noise of diffusion-based WMs. However, optimizing high-dimensional noise is challenging: the optimization must reason about nuanced, scene-dependent target events in generated videos while avoiding out-of-distribution (OOD) noise that yields implausible imaginations. We address this with two complementary objectives: a semantic objective with a Vision-Language Model that provides informative gradients by reasoning about the generated video, and a plausibility objective that prevents the optimized noise from drifting OOD. With state-of-the-art video world models for autonomous driving and robotic manipulation, we show that StressDream effectively steers imaginations toward high-impact yet plausible outcomes specified by text at inference time, such as task failures, enabling robust policy evaluation and improvement by identifying actions whose plausible futures include undesirable outcomes. Video results are available at https://junwon.me/StressDream/.

Coordinated Diffusion: Generating Multi-Agent Behavior Without Multi-Agent Demonstrations

Imitation learning powered by generative models has proven effective for modeling complex single-agent behaviors. However, teaching multi-agent systems, like multiple arms or vehicles, to coordinate through imitation learning is hindered by a fundamental data bottleneck: as the joint state-action space grows exponentially with the number of agents, collecting a sufficient amount of coordinated multi-agent demonstrations becomes extremely costly. In this work, we ask: how can we leverage single-agent demonstration data to learn multi-agent policies? We present Coordinated Diffusion (CoDi), a framework that couples independently trained single-agent diffusion policies through a user-defined multi-agent cost function, without requiring any coordinated demonstrations. We derive a new diffusion-based sampling scheme wherein the diffusion score function decomposes into independent, single-agent pre-trained base policies plus a cost-driven guidance term that coordinates these base policies into cohesive multi-agent behavior. We show that this guidance term can be estimated in a gradient-free manner, making CoDi applicable to black-box, non-differentiable cost functions without additional training. Theoretically and empirically, we analyze the conditions under which this composition can faithfully approximate a target multi-agent behavior. We find a complementary role for demonstration data versus the cost function: single-agent demonstrations must cover the support of the desired multi-agent behavior, while the cost function must promote desired behavior from this product of single-agent policies. Our results in simulation and hardware experiments of a two-arm manipulation task show that CoDi discovers robust coordinated behavior from single-agent data, is more data-efficient than multi-agent baselines, and highlights the importance of joint guidance, base policy support, and cost design.

When to Act, Ask, or Learn: Uncertainty-Aware Policy Steering

Policy steering is an emerging way to adapt robot behaviors at deployment-time: a learned verifier analyzes low-level action samples proposed by a pre-trained policy (e.g., diffusion policy) and selects only those aligned with the task. While Vision-Language Models (VLMs) are promising general-purpose verifiers due to their reasoning capabilities, existing frameworks often assume these models are well-calibrated. In practice, the overconfident judgment from VLM can degrade the steering performance under both high-level semantic uncertainty in task specifications and low-level action uncertainty or incapability of the pre-trained policy. We propose uncertainty-aware policy steering (UPS), a framework that jointly reasons about semantic task uncertainty and low-level action feasibility, and selects an uncertainty resolution strategy: execute a high-confidence action, clarify task ambiguity via natural language queries, or ask for action interventions to correct the low-level policy when it is deemed incapable at the task. We leverage conformal prediction to calibrate the composition of the VLM and the pre-trained base policy, providing statistical assurances that the verifier selects the correct strategy. After collecting interventions during deployment, we employ residual learning to improve the capability of the pre-trained policy, enabling the system to learn continually but with minimal expensive human feedback. We demonstrate our framework through experiments in simulation and on hardware, showing that UPS can disentangle confident, ambiguous, and incapable scenarios and minimizes expensive user interventions compared to uncalibrated baselines and prior human- or robot-gated continual learning approaches. Videos can be found at https://jessie-yuan.github.io/ups/

Adapting by Analogy: OOD Generalization of Visuomotor Policies via Functional Correspondence

End-to-end visuomotor policies trained using behavior cloning have shown a remarkable ability to generate complex, multi-modal low-level robot behaviors. However, at deployment time, these policies still struggle to act reliably when faced with out-of-distribution (OOD) visuals induced by objects, backgrounds, or environment changes. Prior works in interactive imitation learning solicit corrective expert demonstrations under the OOD conditions -- but this can be costly and inefficient. We observe that task success under OOD conditions does not always warrant novel robot behaviors. In-distribution (ID) behaviors can directly be transferred to OOD conditions that share functional similarities with ID conditions. For example, behaviors trained to interact with in-distribution (ID) pens can apply to interacting with a visually-OOD pencil. The key challenge lies in disambiguating which ID observations functionally correspond to the OOD observation for the task at hand. We propose that an expert can provide this OOD-to-ID functional correspondence. Thus, instead of collecting new demonstrations and re-training at every OOD encounter, our method: (1) detects the need for feedback by first checking if current observations are OOD and then identifying whether the most similar training observations show divergent behaviors, (2) solicits functional correspondence feedback to disambiguate between those behaviors, and (3) intervenes on the OOD observations with the functionally corresponding ID observations to perform deployment-time generalization. We validate our method across diverse real-world robotic manipulation tasks with a Franka Panda robotic manipulator. Our results show that test-time functional correspondences can improve the generalization of a vision-based diffusion policy to OOD objects and environment conditions with low feedback.

Uncertainty-aware Latent Safety Filters for Avoiding Out-of-Distribution Failures

Recent advances in generative world models have enabled classical safe control methods, such as Hamilton-Jacobi (HJ) reachability, to generalize to complex robotic systems operating directly from high-dimensional sensor observations. However, obtaining comprehensive coverage of all safety-critical scenarios during world model training is extremely challenging. As a result, latent safety filters built on top of these models may miss novel hazards and even fail to prevent known ones, overconfidently misclassifying risky out-of-distribution (OOD) situations as safe. To address this, we introduce an uncertainty-aware latent safety filter that proactively steers robots away from both known and unseen failures. Our key idea is to use the world model's epistemic uncertainty as a proxy for identifying unseen potential hazards. We propose a principled method to detect OOD world model predictions by calibrating an uncertainty threshold via conformal prediction. By performing reachability analysis in an augmented state space-spanning both the latent representation and the epistemic uncertainty-we synthesize a latent safety filter that can reliably safeguard arbitrary policies from both known and unseen safety hazards. In simulation and hardware experiments on vision-based control tasks with a Franka manipulator, we show that our uncertainty-aware safety filter preemptively detects potential unsafe scenarios and reliably proposes safe, in-distribution actions. Video results can be found on the project website at https://cmu-intentlab.github.io/UNISafe

A Unified Framework for Robots that Influence Humans over Long-Term Interaction

Robot actions influence the decisions of nearby humans. Here influence refers to intentional change: robots influence humans when they shift the human's behavior in a way that helps the robot complete its task. Imagine an autonomous car trying to merge; by proactively nudging into the human's lane, the robot causes human drivers to yield and provide space. Influence is often necessary for seamless interaction. However, if influence is left unregulated and uncontrolled, robots will negatively impact the humans around them. Prior works have begun to address this problem by creating a variety of control algorithms that seek to influence humans. Although these methods are effective in the short-term, they fail to maintain influence over time as the human adapts to the robot's behaviors. In this paper we therefore present an optimization framework that enables robots to purposely regulate their influence over humans across both short-term and long-term interactions. Here the robot maintains its influence by reasoning over a dynamic human model which captures how the robot's current choices will impact the human's future behavior. Our resulting framework serves to unify current approaches: we demonstrate that state-of-the-art methods are simplifications of our underlying formalism. Our framework also provides a principled way to generate influential policies: in the best case the robot exactly solves our framework to find optimal, influential behavior. But when solving this optimization problem becomes impractical, designers can introduce their own simplifications to reach tractable approximations. We experimentally compare our unified framework to state-of-the-art baselines and ablations, and demonstrate across simulations and user studies that this framework is able to successfully influence humans over repeated interactions. See videos of our experiments here: https://youtu.be/nPekTUfUEbo

From Foresight to Forethought: VLM-In-the-Loop Policy Steering via Latent Alignment

While generative robot policies have demonstrated significant potential in learning complex, multimodal behaviors from demonstrations, they still exhibit diverse failures at deployment-time. Policy steering offers an elegant solution to reducing the chance of failure by using an external verifier to select from low-level actions proposed by an imperfect generative policy. Here, one might hope to use a Vision Language Model (VLM) as a verifier, leveraging its open-world reasoning capabilities. However, off-the-shelf VLMs struggle to understand the consequences of low-level robot actions as they are represented fundamentally differently than the text and images the VLM was trained on. In response, we propose FOREWARN, a novel framework to unlock the potential of VLMs as open-vocabulary verifiers for runtime policy steering. Our key idea is to decouple the VLM's burden of predicting action outcomes (foresight) from evaluation (forethought). For foresight, we leverage a latent world model to imagine future latent states given diverse low-level action plans. For forethought, we align the VLM with these predicted latent states to reason about the consequences of actions in its native representation--natural language--and effectively filter proposed plans. We validate our framework across diverse robotic manipulation tasks, demonstrating its ability to bridge representational gaps and provide robust, generalizable policy steering.

Generalizing Safety Beyond Collision-Avoidance via Latent-Space Reachability Analysis

Hamilton-Jacobi (HJ) reachability is a rigorous mathematical framework that enables robots to simultaneously detect unsafe states and generate actions that prevent future failures. While in theory, HJ reachability can synthesize safe controllers for nonlinear systems and nonconvex constraints, in practice, it has been limited to hand-engineered collision-avoidance constraints modeled via low-dimensional state-space representations and first-principles dynamics. In this work, our goal is to generalize safe robot controllers to prevent failures that are hard -- if not impossible -- to write down by hand, but can be intuitively identified from high-dimensional observations: for example, spilling the contents of a bag. We propose Latent Safety Filters, a latent-space generalization of HJ reachability that tractably operates directly on raw observation data (e.g., RGB images) by performing safety analysis in the latent embedding space of a generative world model. This transforms nuanced constraint specification to a classification problem in latent space and enables reasoning about dynamical consequences that are hard to simulate. In simulation and hardware experiments, we use Latent Safety Filters to safeguard arbitrary policies (from generative policies to direct teleoperation) from complex safety hazards, like preventing a Franka Research 3 manipulator from spilling the contents of a bag or toppling cluttered objects.

Your Learned Constraint is Secretly a Backward Reachable Tube

Inverse Constraint Learning (ICL) is the problem of inferring constraints from safe (i.e., constraint-satisfying) demonstrations. The hope is that these inferred constraints can then be used downstream to search for safe policies for new tasks and, potentially, under different dynamics. Our paper explores the question of what mathematical entity ICL recovers. Somewhat surprisingly, we show that both in theory and in practice, ICL recovers the set of states where failure is inevitable, rather than the set of states where failure has already happened. In the language of safe control, this means we recover a backwards reachable tube (BRT) rather than a failure set. In contrast to the failure set, the BRT depends on the dynamics of the data collection system. We discuss the implications of the dynamics-conditionedness of the recovered constraint on both the sample-efficiency of policy search and the transferability of learned constraints.