Task Success is not Enough: Investigating the Use of Video-Language Models as Behavior Critics for Catching Undesirable Agent Behaviors

• URL: http://arxiv.org/abs/2402.04210v2
• Date: Sun, 11 Aug 2024 07:13:50 GMT
• Title: Task Success is not Enough: Investigating the Use of Video-Language Models as Behavior Critics for Catching Undesirable Agent Behaviors
• Abstract summary: Large-scale generative models are useful for sampling meaningful candidate solutions, yet they often overlook task constraints and user preferences. In the context of embodied AI, verification often solely involves assessing whether goal conditions specified in the instructions have been met. Given the scope of robot tasks, it is infeasible to construct scripted verifiers akin to those used for explicit-knowledge tasks like Go. This begs the question: can we use large vision and language models (VLMs), which are approximately omniscient, as scalable Behavior Critics to catch undesirable robot behaviors in videos?
• Score: 22.728139463769896
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Large-scale generative models are shown to be useful for sampling meaningful candidate solutions, yet they often overlook task constraints and user preferences. Their full power is better harnessed when the models are coupled with external verifiers and the final solutions are derived iteratively or progressively according to the verification feedback. In the context of embodied AI, verification often solely involves assessing whether goal conditions specified in the instructions have been met. Nonetheless, for these agents to be seamlessly integrated into daily life, it is crucial to account for a broader range of constraints and preferences beyond bare task success (e.g., a robot should grasp bread with care to avoid significant deformations). However, given the unbounded scope of robot tasks, it is infeasible to construct scripted verifiers akin to those used for explicit-knowledge tasks like the game of Go and theorem proving. This begs the question: when no sound verifier is available, can we use large vision and language models (VLMs), which are approximately omniscient, as scalable Behavior Critics to catch undesirable robot behaviors in videos? To answer this, we first construct a benchmark that contains diverse cases of goal-reaching yet undesirable robot policies. Then, we comprehensively evaluate VLM critics to gain a deeper understanding of their strengths and failure modes. Based on the evaluation, we provide guidelines on how to effectively utilize VLM critiques and showcase a practical way to integrate the feedback into an iterative process of policy refinement. The dataset and codebase are released at: https://guansuns.github.io/pages/vlm-critic.

Related papers

• Beyond the Traceback: Using LLMs for Adaptive Explanations of Programming Errors [52.619793817715326]
  We present a multi-stage crowdsourced study evaluating skill-targeted, LLM-generated Python error messages.<n>We show that while LLM-rewritten messages significantly improved subjective evaluations, these perceived gains did not translate into statistically significant improvements in objective debug performance.<n>This highlights a critical human-AI complementarity gap: explanations that feel better to users do not necessarily make them more effective debuggers.
  arXiv  Detail & Related papers  (2026-08-21T09:14:28Z)
• RoboTALES: Learning Reasoning-Guided Robot Policies via Task-Aligned Simulated Futures [51.85284871989367]
  We propose RoboTALES, a framework that learns task-aligned simulated futures and uses them to train robot policies.<n>Our approach introduces two key innovations: (1) a hierarchical LLM-based planner that breaks complex tasks into a sequence of subgoals to guide the model's imagination; and (2) a VLM-based critic that evaluates these imagined'' futures and uses reward-based feedback to keep the model's internal representations focused on the goal.
  arXiv  Detail & Related papers  (2026-07-07T08:58:29Z)
• Employing Vision-Language Models for Face Image Quality Assessment [0.023141219541346198]
  Face Image Quality Assessment (FIQA) is a crucial control step in biometric pipelines.<n>State-of-the-art FIQA methods achieve high utility but typically operate as "black boxes"<n>We investigate the potential of off-the-shelf Vision-Language Models (VLMs) to bridge this gap by performing FIQA in a zero-shot setting.
  arXiv  Detail & Related papers  (2026-05-17T14:57:52Z)
• LIBERO-PRO: Towards Robust and Fair Evaluation of Vision-Language-Action Models Beyond Memorization [33.562814882942]
  LIBERO has emerged as a widely adopted benchmark for evaluating Vision-Language-Action (VLA) models.<n>We introduce LIBERO-PRO, an extended LIBERO benchmark that systematically evaluates model performance under reasonable perturbations.<n> Experimental results reveal that, although existing models achieve over 90% accuracy under the standard LIBERO evaluation, their performance collapses to 0.0% under our generalized setting.
  arXiv  Detail & Related papers  (2025-10-04T14:56:40Z)
• Critique to Verify: Accurate and Honest Test-Time Scaling with RL-Trained Verifiers [63.99316853136304]
  Mirror-Critique is a framework that trains a verifier with informative critiques.<n>We deploy a small instruction-tuned model to synthesize high-quality critique data.<n>The resulting Mirror-Verifier is deployed to evaluate candidate solutions by generating multiple critiques per solution.
  arXiv  Detail & Related papers  (2025-09-27T06:50:24Z)
• Measuring Epistemic Humility in Multimodal Large Language Models [17.490955813494693]
  We present HumbleBench, a new hallucination benchmark designed to evaluate MLLMs' ability to reject plausible but incorrect answers.<n>We leverage fine-grained scene graph annotations to extract ground-truth entities and relations, and prompt GPT-4-Turbo to generate multiple-choice questions.<n>HumbleBench fills a key gap in current evaluation suites, providing a more realistic measure of MLLM reliability in safety-critical settings.
  arXiv  Detail & Related papers  (2025-09-11T17:54:00Z)
• Do What? Teaching Vision-Language-Action Models to Reject the Impossible [53.40183895299108]
  Vision-Language-Action (VLA) models have demonstrated strong performance on a range of robotic tasks.<n>We propose Instruct-Verify-and-Act (IVA), a framework that detects when an instruction cannot be executed due to a false premise.<n>Our experiments show that IVA improves false premise detection accuracy by 97.56% over baselines.
  arXiv  Detail & Related papers  (2025-08-22T10:54:33Z)
• Let's Think in Two Steps: Mitigating Agreement Bias in MLLMs with Self-Grounded Verification [17.67273082468732]
  Verifiers -- functions assigning rewards to agent behavior -- have been key for AI progress in domains like math and board games.<n>We evaluate Multimodal Large Language Models (MLLMs) as verifiers of agent trajectories across web navigation, computer use, and robotic manipulation.<n>We propose Self-Grounded Verification (SGV), a lightweight method that enables more effective use of MLLMs' knowledge and reasoning.
  arXiv  Detail & Related papers  (2025-07-15T18:50:29Z)
• GoalLadder: Incremental Goal Discovery with Vision-Language Models [38.35578010611503]
  We propose a novel method to train RL agents from a single language instruction in visual environments.<n>GoalLadder works by incrementally discovering states that bring the agent closer to completing a task specified in natural language.<n>Unlike prior work, GoalLadder does not trust VLM's feedback completely; instead, it uses it to rank potential goal states using an ELO-based rating system.
  arXiv  Detail & Related papers  (2025-06-19T15:28:27Z)
• From Intention to Execution: Probing the Generalization Boundaries of Vision-Language-Action Models [5.660635614478238]
  Vision-Language-Action (VLA) models promise to produce versatile, "generalist" robot policies.<n>Traditional imitation learning benchmarks are unsuitable due to the lack of language instructions.<n>We introduce a unified suite of 50 simulation-based tasks across 10 subcategories spanning language instruction, vision, and objects.
  arXiv  Detail & Related papers  (2025-06-11T16:52:18Z)
• Reasoning Multimodal Large Language Model: Data Contamination and Dynamic Evaluation [9.434966074326056]
  Multimodal Large Language Models (MLLMs) show impressive vision-language benchmark performance, yet growing concerns about data contamination risk masking true generalization.<n>We propose a novel dynamic evaluation framework to rigorously assess MLLM generalization, moving beyond static benchmarks.<n>We demonstrate that fine-tuning on simulated test data (extreme contamination) drastically sharpens task-specific performance but harms overall generalization.
  arXiv  Detail & Related papers  (2025-06-08T15:52:38Z)
• Maximizing Signal in Human-Model Preference Alignment [0.0]
  This paper argues that in cases in which end users need to agree with the decisions made by ML models, models should be trained and evaluated on data that represent their preferences. We show that noise in labeling disagreement can be minimized by adhering to proven methodological best practices.
  arXiv  Detail & Related papers  (2025-03-06T19:10:57Z)
• SPARC: Score Prompting and Adaptive Fusion for Zero-Shot Multi-Label Recognition in Vision-Language Models [74.40683913645731]
  Zero-shot multi-label recognition (MLR) with Vision-Language Models (VLMs) faces significant challenges without training data, model tuning, or architectural modifications. Our work proposes a novel solution treating VLMs as black boxes, leveraging scores without training data or ground truth. Analysis of these prompt scores reveals VLM biases and AND''/OR' signal ambiguities, notably that maximum scores are surprisingly suboptimal compared to second-highest scores.
  arXiv  Detail & Related papers  (2025-02-24T07:15:05Z)
• Interactive Agents to Overcome Ambiguity in Software Engineering [61.40183840499932]
  AI agents are increasingly being deployed to automate tasks, often based on ambiguous and underspecified user instructions. Making unwarranted assumptions and failing to ask clarifying questions can lead to suboptimal outcomes. We study the ability of LLM agents to handle ambiguous instructions in interactive code generation settings by evaluating proprietary and open-weight models on their performance.
  arXiv  Detail & Related papers  (2025-02-18T17:12:26Z)
• Enhancing LLM Reasoning via Critique Models with Test-Time and Training-Time Supervision [120.40788744292739]
  We propose a two-player paradigm that separates the roles of reasoning and critique models. We first propose AutoMathCritique, an automated and scalable framework for collecting critique data. We demonstrate that the critique models consistently improve the actor's performance on difficult queries at test-time.
  arXiv  Detail & Related papers  (2024-11-25T17:11:54Z)
• Vision Language Models are In-Context Value Learners [89.29486557646624]
  We present Generative Value Learning (GVL), a universal value function estimator that leverages the world knowledge embedded in vision-language models (VLMs) to predict task progress. Without any robot or task specific training, GVL can in-context zero-shot and few-shot predict effective values for more than 300 distinct real-world tasks.
  arXiv  Detail & Related papers  (2024-11-07T09:17:50Z)
• A Closer Look at Machine Unlearning for Large Language Models [46.245404272612795]
  Large language models (LLMs) may memorize sensitive or copyrighted content, raising privacy and legal concerns. We discuss several issues in machine unlearning for LLMs and provide our insights on possible approaches.
  arXiv  Detail & Related papers  (2024-10-10T16:56:05Z)
• No Regrets: Investigating and Improving Regret Approximations for Curriculum Discovery [53.08822154199948]
  Unsupervised Environment Design (UED) methods have gained recent attention as their adaptive curricula promise to enable agents to be robust to in- and out-of-distribution tasks. This work investigates how existing UED methods select training environments, focusing on task prioritisation metrics. We develop a method that directly trains on scenarios with high learnability.
  arXiv  Detail & Related papers  (2024-08-27T14:31:54Z)
• Robotic Control via Embodied Chain-of-Thought Reasoning [86.6680905262442]
  Key limitation of learned robot control policies is their inability to generalize outside their training data. Recent works on vision-language-action models (VLAs) have shown that the use of large, internet pre-trained vision-language models can substantially improve their robustness and generalization ability. We introduce Embodied Chain-of-Thought Reasoning (ECoT) for VLAs, in which we train VLAs to perform multiple steps of reasoning about plans, sub-tasks, motions, and visually grounded features before predicting the robot action.
  arXiv  Detail & Related papers  (2024-07-11T17:31:01Z)
• Explore until Confident: Efficient Exploration for Embodied Question Answering [32.27111287314288]
  We leverage the strong semantic reasoning capabilities of large vision-language models to efficiently explore and answer questions. We propose a method that first builds a semantic map of the scene based on depth information and via visual prompting of a VLM. Next, we use conformal prediction to calibrate the VLM's question answering confidence, allowing the robot to know when to stop exploration.
  arXiv  Detail & Related papers  (2024-03-23T22:04:03Z)
• Interactive Planning Using Large Language Models for Partially Observable Robotics Tasks [54.60571399091711]
  Large Language Models (LLMs) have achieved impressive results in creating robotic agents for performing open vocabulary tasks. We present an interactive planning technique for partially observable tasks using LLMs.
  arXiv  Detail & Related papers  (2023-12-11T22:54:44Z)
• Robust Tests in Online Decision-Making [3.867363075280544]
  Bandit algorithms are widely used in sequential decision problems to maximize the cumulative reward. In mobile health, the goal is to promote the user's health through personalized interventions based on user specific information acquired through wearable devices. We propose a modified actor-critic algorithm which is robust to critic misspecification and derive a novel testing procedure for the actor parameters.
  arXiv  Detail & Related papers  (2022-08-21T06:50:45Z)
• Adversarial GLUE: A Multi-Task Benchmark for Robustness Evaluation of Language Models [86.02610674750345]
  Adversarial GLUE (AdvGLUE) is a new multi-task benchmark to explore and evaluate the vulnerabilities of modern large-scale language models under various types of adversarial attacks. We apply 14 adversarial attack methods to GLUE tasks to construct AdvGLUE, which is further validated by humans for reliable annotations. All the language models and robust training methods we tested perform poorly on AdvGLUE, with scores lagging far behind the benign accuracy.
  arXiv  Detail & Related papers  (2021-11-04T12:59:55Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.