Guardrailed Meta-Agent Loops: Stress-Testing Policy Pinning, Budget Bounds, and Crash Recovery

Qinzhen Ma; Jialin Wu

Home/Archive/arXiv

arXiv · 2609.12216

Guardrailed Meta-Agent Loops: Stress-Testing Policy Pinning, Budget Bounds, and Crash Recovery

Qinzhen Ma·Jialin Wu

View original source ↗Read source document ↗Export metadata ↓

Abstract

Self-improving agent workflows create an audit problem when the same controller can change both its behavior and the conditions under which that behavior is judged. We present GuardrailLoop, a simulation-based testbed that makes three operational contracts jointly testable: preservation of human-defined policy, compute accounting at every recorded execution prefix, and recovery of a specified scientific state after crashes. A hash-pinned policy fixes goals, scope, evaluation identity, budget, and release conditions; machine-directed evolution is restricted to a code-owned feature catalog and bounded knobs. The contribution is an executable boundary and an evaluation protocol that separates useful adaptation, state recovery, and repeated execution. In a paired 50-seed 2 x 2 study, round-stage growth changes target attainment by +1.00 and restricted mean compute to target by -56.97 simulated GPU-hours (95% paired-bootstrap interval [-58.91,-54.70]); idle growth has zero measured utility effect. Across 240 enumerated crash injections, all runs recover the defined outcome, but only 210 preserve the normalized trace: 30 pre-commit crashes repeat a planner call. Resource-drift, kill-switch, integrity, and output-guard matrices satisfy their specified checks. These findings show why successful outcome recovery is insufficient evidence of exactly-once execution. They establish conformance within one calibrated deterministic testbed, rather than general safety or real-world self-improvement.

Explore related subjects

cs.RO

Keep this discovery

Explore connections, maps & timelines

BibTeX RIS

Qinzhen Ma, Jialin Wu. 2026-09-10. Guardrailed Meta-Agent Loops: Stress-Testing Policy Pinning, Budget Bounds, and Crash Recovery. https://arxiv.org/abs/2609.12216

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Robotic Tele-Operation for Upper Aerodigestive Tract Microsurgery: System Design and Validation

Upper aerodigestive tract (UADT) treatments frequently employ transoral laser microsurgery (TLM) for procedures such as the removal of tumors or polyps. In TLM, a laser beam is used to cut target tissue, while forceps are employed to grasp, manipulate, and stabilize tissue within the UADT. Although TLM systems may rely on different technologies and interfaces, forceps manipulation is still predominantly performed manually, introducing limitations in ergonomics, precision, and controllability. This paper proposes a novel robotic system for tissue manipulation in UADT procedures, based on a novel end-effector designed for forceps control. The system is integrated within a teleoperation framework that employs a robotic manipulator with a programmed remote center of motion (RCM), enabling precise and constrained instrument motion while improving surgeon ergonomics. The proposed approach is validated through two experimental studies and a dedicated usability evaluation, demonstrating its effectiveness and suitability for UADT surgical applications.

cs.RO↗

arXiv2026

HERMES: A Holistic End-to-End Risk-Aware Multimodal Embodied System with Vision-Language Models for Long-Tail Autonomous Driving

End-to-end autonomous driving models increasingly benefit from large vision-language models for semantic understanding, yet safe and reliable planning under long-tail conditions remains challenging, particularly in mixed-traffic environments involving heterogeneous road users and rare safety-critical interactions. This paper proposes HERMES, a holistic risk-aware end-to-end multimodal driving framework that explicitly incorporates long-tail semantic knowledge into trajectory planning. HERMES employs a foundation-model-assisted annotation pipeline to construct structured Long-Tail Scene Context and Long-Tail Planning Context, capturing hazard-centric scene information, maneuver intent, and risk-aware planning guidance. A Tri-Modal Driving Module then integrates multi-view visual observations, historical ego-motion, and long-tail semantic instructions through intent- and risk-aware conditioning for trajectory generation. Extensive experiments on a large-scale real-world long-tail driving benchmark demonstrate consistent improvements over representative recent baselines in overall planning performance and across diverse safety-critical scenarios. Ablation studies further validate the effectiveness and complementary roles of the major components within HERMES.

cs.RO↗

arXiv2026

EgoPush: Egocentric Multi-Object Rearrangement for Mobile Robots via Constrained Teacher Observability

Humans rearrange objects in cluttered environments using egocentric perception, actively moving to keep task-relevant spatial cues in view. Mobile robots have not matched this: rearrangement is usually built on a global pose estimate or a map, which is exactly what a robot carrying one camera lacks, while pushing keeps changing the scene it would have to be built from. We present EgoPush, which pushes objects into anchor-relative formations from onboard RGB-D alone, with no global localization, external tracking, or map at deployment, and transfers zero-shot to a TurtleBot in controlled and visually cluttered scenes. What makes this learnable turns out to be a property of the teacher rather than of the student: three privileged teachers trained with identical rewards, architecture, and hyperparameters all exceed $98\%$ success, yet their distilled egocentric students reach $0\%$, $54.8\%$, and $87.3\%$, the only variable being the teacher's observation function. EgoPush therefore trains the teacher under egocentric observability constraints, restricting it to visibility-limited cues and revealing target references only when the anchor is centrally visible, so that its supervision is recoverable by a depth-based student distilled online. Making the teacher trainable in the first place needs two further pieces: a role-grouped object-centric interface shared by teacher and student, and stage-wise temporally decayed rewards for long-horizon credit assignment. Videos, the playable task, and code are available at https://ai4ce.github.io/EgoPush/.

cs.RO↗