Steve-Evolving: Open-World Embodied Self-Evolution via Fine-Grained Diagnosis and Dual-Track Knowledge Distillation

• URL: http://arxiv.org/abs/2603.13131v1
• Date: Fri, 13 Mar 2026 16:23:34 GMT
• Title: Steve-Evolving: Open-World Embodied Self-Evolution via Fine-Grained Diagnosis and Dual-Track Knowledge Distillation
• Abstract summary: We present a non-parametric self-evolving framework that tightly couples fine-grained execution diagnosis with dual-track knowledge distillation in a closed loop.<n>The method follows three phases: Experience Anchoring, Experience Distillation, and KnowledgeDriven ClosedLoopLoop.<n> Experiments on the long-horizon suite of Minecraft MCU demonstrate consistent trajectories over static-retrievals.
• Score: 8.148658798211313
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Open-world embodied agents must solve long-horizon tasks where the main bottleneck is not single-step planning quality but how interaction experience is organized and evolved. To this end, we present Steve-Evolving, a non-parametric self-evolving framework that tightly couples fine-grained execution diagnosis with dual-track knowledge distillation in a closed loop. The method follows three phases: Experience Anchoring, Experience Distillation, and Knowledge-Driven Closed-Loop Control. In detail, Experience Anchoring solidifies each subgoal attempt into a structured experience tuple with a fixed schema (pre-state, action, diagnosis-result, and post-state) and organizes it in a three-tier experience space with multi-dimensional indices (e.g., condition signatures, spatial hashing, and semantic tags) plus rolling summarization for efficient and auditable recall. To ensure sufficient information density for attribution, the execution layer provides compositional diagnosis signals beyond binary outcomes, including state-difference summaries, enumerated failure causes, continuous indicators, and stagnation/loop detection. Moreover, successful trajectories of Experience Distillation are generalized into reusable skills with explicit preconditions and verification criteria, while failures are distilled into executable guardrails that capture root causes and forbid risky operations at both subgoal and task granularities. Besides, Knowledge-Driven Closed-Loop Control retrieved skills and guardrails are injected into an LLM planner, and diagnosis-triggered local replanning updates the active constraints online, forming a continual evolution process without any model parameter updates. Experiments on the long-horizon suite of Minecraft MCU demonstrate consistent improvements over static-retrieval baselines.

Related papers

• HarnessEvolve: Learning from Reference Trajectories for Reliable Agent Self-Evolution [11.431479238875712]
  We introduce HarnessEvolve, a self-evolving framework that learns from reference trajectories to achieve reliable agent self-evolution.<n>We conduct extensive experiments on several benchmarks spanning open-domain and enterprise scenarios, using different models and agent frameworks.<n>Results demonstrate that HarnessEvolve consistently outperforms state-of-the-art baselines across all benchmarks and settings.
  arXiv  Detail & Related papers  (2026-09-01T07:31:18Z)
• Shift & Drift: A Zero-Shot Benchmark for Generalizable and Robust Autonomous Driving Motion Planning [60.888463449435484]
  We present Shift & Drift, a novel dual-track benchmark designed to rigorously stress-test motion planners across two critical axes of distribution shift.<n>We show that imitation learning methods achieve high scores in ID benchmarks, but suffer from persistent drift when subjected to temporally correlated actuation noise.<n>Our findings reveal an empirical trade-off between imitation fidelity and closed-loop resilience, providing the community with a rigorous benchmark to evaluate progress toward reliable deployment.
  arXiv  Detail & Related papers  (2026-07-08T18:23:48Z)
• StainFlow: Entity-Stain Tracking and Evidence Linking for Process Rewards in GUI Agents [67.03593791535786]
  Reinforcement Learning (RL) has become a promising approach for improving GUI Agents in long-horizon, digital environments.<n>To mitigate this issue, recent studies introduce Process Reward Models (PRMs)<n>PRMs provide finer-grained training feedback through global milestone verification or local step-level evaluation.<n>We propose StainFlow, an entity-stain-flow process reward model for GUI Agents.
  arXiv  Detail & Related papers  (2026-06-05T08:17:28Z)
• AI Knows What's Wrong But Cannot Fix It: Helicoid Dynamics in Frontier LLMs Under High-Stakes Decisions [51.56484100374058]
  Helicoid dynamics is the name given to a specific failure regime in that second domain.<n>System engages competently, drifts into error, accurately names what went wrong, then reproduces the same pattern at a higher level of sophistication.<n>This prospective case series documents that regime across seven leading systems.
  arXiv  Detail & Related papers  (2026-03-12T05:25:49Z)
• LHAW: Controllable Underspecification for Long-Horizon Tasks [8.46227536869596]
  We introduce LHAW (Long-Horizon Augmenteds), a modular, dataset-agnostic synthetic pipeline that transforms any well-specified task into controllable underspecified variants.<n>Unlike approaches that rely on LLM predictions of ambiguity, LHAW validates variants through empirical agent trials, classifying them as outcome-critical, divergent, or benign based on observed terminal state divergence.<n>We release 285 task variants from TheAgentCompany, SWE-Bench Pro and MCP-Atlas, measuring how current agents detect, reason about, and resolve underspecification across ambiguous settings.
  arXiv  Detail & Related papers  (2026-02-11T04:49:50Z)
• daVinci-Agency: Unlocking Long-Horizon Agency Data-Efficiently [35.39097522391409]
  Large Language Models (LLMs) excel at short-term tasks, scaling them to long-horizon agentic synthesis remains challenging.<n>We propose daVinci-Agency, which systematically mines structured supervision from chain-of-PRs.<n>DaVinci-Agency's PR-grounded structure inherently preserves the causal dependencies and iterative refinements essential for teaching persistent goal-cycle behavior.
  arXiv  Detail & Related papers  (2026-02-02T13:23:39Z)
• Hierarchical Vision Language Action Model Using Success and Failure Demonstrations [60.82332413442677]
  We introduce VINE, a hierarchical vision-language-action model that separates high-level reasoning from low-level control.<n>System 2 performs feasibility-guided tree search over a 2D scene-graph abstraction.<n>System 1 executes low-level actions without modifying the agent's core skills.
  arXiv  Detail & Related papers  (2025-12-03T15:58:38Z)
• CKDA: Cross-modality Knowledge Disentanglement and Alignment for Visible-Infrared Lifelong Person Re-identification [77.07028925223383]
  Lifelong person Re-IDentification aims to match the same person employing continuously collected individual data from different scenarios.<n>To achieve continuous all-day person matching across day and night, Visible-Infrared Lifelong person Re-IDentification (VI-LReID) focuses on sequential training on data from visible and infrared modalities.<n>Existing methods typically exploit cross-modal knowledge distillation to alleviate the catastrophic forgetting of old knowledge.
  arXiv  Detail & Related papers  (2025-11-19T01:30:29Z)
• Building a Foundational Guardrail for General Agentic Systems via Synthetic Data [76.18834864749606]
  LLM agents can plan multi-step tasks, intervening at the planning stage-before any action is executed-is often the safest way to prevent harm.<n>Existing guardrails mostly operate post-execution, which is difficult to scale and leaves little room for controllable supervision at the plan level.<n>We introduce AuraGen, a controllable engine that synthesizes benign trajectories, injects category-labeled risks with difficulty, and filters outputs via an automated reward model.
  arXiv  Detail & Related papers  (2025-10-10T18:42:32Z)
• Alignment Tipping Process: How Self-Evolution Pushes LLM Agents Off the Rails [103.05296856071931]
  We identify the Alignment Tipping Process (ATP), a critical post-deployment risk unique to self-evolving Large Language Model (LLM) agents.<n>ATP arises when continual interaction drives agents to abandon alignment constraints established during training in favor of reinforced, self-interested strategies.<n>Our experiments show that alignment benefits erode rapidly under self-evolution, with initially aligned models converging toward unaligned states.
  arXiv  Detail & Related papers  (2025-10-06T14:48:39Z)
• Mutual Information Tracks Policy Coherence in Reinforcement Learning [0.0]
  Reinforcement Learning (RL) agents face degradation from sensor faults, actuator wear, and environmental shifts.<n>We present an information-theoretic framework that reveals both the fundamental dynamics of RL and provides practical methods for diagnosing deployment-time anomalies.
  arXiv  Detail & Related papers  (2025-09-12T17:24:20Z)
• Endo-TTAP: Robust Endoscopic Tissue Tracking via Multi-Facet Guided Attention and Hybrid Flow-point Supervision [3.290418382279656]
  Endo-TTAP is a novel framework for tissue point tracking in endoscopic videos.<n>MFGA module synergizes multi-scale flow dynamics, DINOv2 semantic embeddings, and explicit motion patterns to jointly predict point positions.<n> Stage I utilizes synthetic data with optical flow ground truth for uncertainty-occlusion regularization.<n> Stage II combines unsupervised flow consistency and semi-supervised learning with refined pseudo-labels from off-the-shelf trackers.
  arXiv  Detail & Related papers  (2025-03-28T13:00:07Z)
• Weakly-Supervised Spatio-Temporal Anomaly Detection in Surveillance Video [128.41392860714635]
  We introduce Weakly-Supervised Snoma-Temporally Detection (WSSTAD) in surveillance video. WSSTAD aims to localize a-temporal tube (i.e. sequence of bounding boxes at consecutive times) that encloses abnormal event. We propose a dual-branch network which takes as input proposals with multi-granularities in both spatial-temporal domains.
  arXiv  Detail & Related papers  (2021-08-09T06:11:14Z)

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.