arXiv:2610.09115v1 Announce Type: cross Abstract: Steering an LLM agent means deciding whether to correct it, at which step, and with which mechanism. Uncertainty is often used to decide when to corr...
arXiv:2609.22582v1 Announce Type: new Abstract: End-to-end and vision-language-action (VLA) driving policies are compared by leaderboard rank, but a rank reports an outcome, not the behaviour behind...
arXiv:2610.31835v1 Announce Type: new Abstract: Temporary traffic control redirects vehicles through work zones, requiring automated vehicles to connect an interpretation of the visible devices to a...
arXiv:2606. 29898v1 Announce Type: cross Abstract: Real-world evaluation is the gold standard for robot policies because it tests them against the physical conditions and deployment challenges they are ultimately designed to handle.
Intent misinterpretation during vehicle interactions causes recurring planning failures. We study a decision layer in which a language-guided intent module reads structured descriptors, computes a smo...
arXiv:2610.18614v1 Announce Type: new Abstract: Controlling an acting LLM agent means deciding, at each step of a long-horizon task, whether to trust its next action, abstain, or hand off to a human....
The paper introduces RACE, a rollout‑free metric for selecting robot policy checkpoints that uses held‑out successful trajectories to calibrate action deviations against state‑conditional thresholds. By aggregating these calibrated scores with a running product, RACE reduces the influence of later reference predictions when early deviations occur, and averages the cumulative scores over time and across trajectories to produce a checkpoint score. Experiments in simulation and on real robots show that RACE outperforms baseline methods and its rankings correlate with deployment success across various policy backbones and tasks.
TRACE (TRAjectory-Contrastive Evolution) is a self‑evolving skill bank that improves the consistency and limit‑awareness of large‑language‑model agents without changing the model weights. By iteratively refining modular skills based on successful and failed trajectories, TRACE raises consistent performance (Pass^3) on the CAR‑bench in‑car assistant tasks from 59.9 % to 94.5 % on GPT‑5.5 and achieves first place on the hidden set with GPT‑5.6‑Sol. The approach demonstrates that a skill‑based, self‑evolution loop can convert a model’s potential into stable, reliable behavior.
The paper investigates whether human or LLM‑judge‑identified ‘first mistake’ points truly indicate failure causes in LLM agents. By reconstructing intermediate states and computing a prefix value V(h_t) for each step, the authors identify a certified point of no return (PNR) and compare it to judge‑labeled steps. Their experiments on 150 SWE‑bench tasks show that judge labels often miss the decisive step, while restarting at the measured PNR improves success rates from 0.44 to 0.57.
arXiv:2602. 11619v2 Announce Type: replace Abstract: Running the same LLM agent on identical inputs yields 2.
The paper introduces Traverse, a benchmark of 2,518 agent trajectories and 6,967 annotated mistakes across software engineering, computer use, and science tasks, revealing that failures often go unrecovered and can cause irreversible harm before a run is deemed successful. It shows that human judges struggle to detect the first mistake in most runs, while a 4‑billion‑parameter verifier called Scout can locate failures more effectively and improve task success when used to select among candidate runs. The study demonstrates that making failure detection inexpensive and reliable can enable long‑horizon agents to learn from their own mistakes and increase trustworthiness in autonomous AI.
arXiv:2610.09294v1 Announce Type: cross Abstract: Rapid progress in AI agents has brought growing attention to agent safety, with extensive evaluation focused on digital environments. As agents move...