TinyAstral introduces a two‑stage method to transfer improvements from a large teacher robot agent to a compact student agent. First, the teacher’s harness—its tools, feedback, memory, and recovery logic—is recursively self‑improved by LLM agents, then the enhanced harness is frozen and used to distill a small 4B student that matches the teacher’s performance. The distilled student achieves near‑teacher success rates, outperforms larger open‑source agents, and runs with roughly one‑fifth the computational cost.
arXiv:2609.24974v1 Announce Type: cross Abstract: Agent harnesses, the external systems that mediate model-environment interaction, can substantially improve agent performance, but their gains remain...
RoboArchitect automates the design of agent harnesses that orchestrate frozen robot policies, vision tools, and direct arm commands. By observing rollouts, diagnosing failures, and iteratively revising the harness, it improves success rates from 9.7% to 43.3% on unseen scenes and from 12.7% to 40.2% on unseen tasks without updating model weights. The evolved harness also develops persistent memory and intuitive control strategies, outperforming finetuned policies especially when the benchmark diverges from the policy’s training data.
arXiv:2610.35817v1 Announce Type: cross Abstract: A robot policy needs to understand what a task is asking for in a scene, and how to move the robot to accomplish that task. To achieve these two skil...
Teach-and-Grow Learning (TGL) is an agent-centered architecture that transforms a few successful demonstrations into reusable Skill Blocks, enabling a robot to compose, execute, and revise behaviors in new scenes without task-specific policy retraining. The system maintains a Skill Library and structured Experience Memory to capture successes, failures, and repairs, allowing persistent reuse and agent-directed adaptation. Evaluation on the LIBERO benchmark shows state-of-the-art performance, and the authors propose a scaling-law hypothesis suggesting that accumulated reusable experience reduces future-task error and teaching demand following a power-law trend.
arXiv:2610.18792v1 Announce Type: cross Abstract: Frontier language-model agents can now improve robots on their own: they write control code, test it, and keep what works as skills and memories for...
The paper introduces Regularized Recursive Self-Improvement of Agent Harnesses (RRSI), a method that applies regularization principles to the iterative editing of an LLM agent’s harness—prompts, control flow, tooling, memory, and context management. RRSI limits the number of edits per candidate, encourages novel trajectories, and uses a critic and pruner to filter out benchmark‑specific or ineffective changes, thereby favoring reusable agent mechanisms. Experiments on eight benchmarks show RRSI improves performance by up to 14.1 points on the training split and 4.7 points on out‑of‑distribution tests, while reducing policy token usage by 30% compared to unregularized evolution.
arXiv:2609.39304v1 Announce Type: cross Abstract: When an off-the-shelf coding agent is used directly as a robot policy, observing a browser-based 3D interface through screenshots and acting by posin...
arXiv:2610.20474v1 Announce Type: new Abstract: Pretrained vision-language-action models (VLAs) often fail midway through unseen long-horizon tasks. Recursive self-improvement (RSI) offers a remedy,...
The paper introduces Harness-Aware Distillation (HAD), a method for training smaller language model agents that preserves the surrounding harness—software managing context, tools, and feedback—while focusing distillation on the teacher’s contributions beyond the harness. HAD combines an action preference that contrasts teacher actions with and without harness information, and a validity check that filters out contradictory preference pairs. Experiments on long-horizon agent benchmarks show that HAD outperforms standard on‑policy distillation, reducing unproductive loops and improving error recovery without requiring task rewards or future information.
Robot foundation models provide strong visuomotor control, yet their performance can degrade when object positions or task instructions change. Further improvements often require post-training on subs...
OmniBrain-0 is a lightweight, unified embodied agent that coordinates manipulation and navigation through a shared high‑level decision model. It integrates task understanding, planning, progress tracking, and feedback‑driven recovery, and is trained via embodied pretraining, domain‑specialized agent distillation, and multi‑teacher on‑policy distillation. The framework also introduces Rubric‑Guided Recursive Self‑Improvement, a data flywheel that refines rubrics and trajectories for continual domain‑expert updates, enabling efficient transfer and consolidation of embodied capabilities across multiple benchmarks and real‑world tasks.
arXiv:2608. 08466v1 Announce Type: new Abstract: Modern LLM agents are often improved by modifying prompts, tools, or workflows manually, while the executable scaffold surrounding the model---the \emph{harness}---is typically treated as a fixed artifact after deployment.