Searcharxiv⌕ Search
Home/Archive/arXiv

arXiv · 2609.32416

RE-0: Verified Recursive Improvement of Embodied Code-as-Policy Agents through Local On-Policy Distillation

Jiawei Zhang·Xiangrong Zhang·Rui Song·Huanbin Zhou·Chengye Song·Hongzhou Wang
View original source ↗Read source document ↗Export metadata ↓

Abstract

Code-as-Policy agents accomplish long-horizon embodied tasks by generating and executing code, yet continually improving them with teachers that are stronger but not globally reliable remains a key challenge. Existing distillation methods typically treat the teacher's complete behavior as the supervision target and thus misassign training credit on states where the teacher fails. We propose RE-0, a recursively verified policy improvement framework: rather than assuming that the teacher globally outperforms the student, RE-0 requests local corrections from the teacher on the student's own failure histories and checks in the environment whether each correction is genuinely beneficial; verified corrections yield immediate improvement. Building on this, we propose RE-OPD, which turns verified interventions into supervision for on-policy distillation. Only counterfactually verified teacher interventions provide distribution-level supervision, weighted by their measured local benefit, and the improvement they induce is projected back into the standalone student, so both where supervision is applied and how much credit the teacher receives co-evolve with the student policy. We further prove that the student's per-round gain is lower-bounded by its verified intervention gain up to verification and projection error terms. Experiments across multiple Code-as-Policy embodied tasks show that RE-0 improves both teacher-assisted execution and the standalone student, and generalizes to novel robots and scenes.

Explore related subjects

cs.ROcs.AIcs.LG

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Jiawei Zhang, Xiangrong Zhang, Rui Song, Huanbin Zhou, Chengye Song, Hongzhou Wang. 2026-09-26. RE-0: Verified Recursive Improvement of Embodied Code-as-Policy Agents through Local On-Policy Distillation. https://arxiv.org/abs/2609.32416

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

KEEP EXPLORING

Related papers

arXiv2026

Fading Expert Guidance: Bridging Model-Based and Learning-Based Control for Abortable Autonomous Overtaking

Overtaking on two-lane roads is a safety-critical decision-making problem for autonomous vehicles, since oncoming traffic may force the ego vehicle to abort and merge back to its original lane. Deep reinforcement learning (DRL) is promising for such continuous-control tasks, but it requires substantial interaction data and may produce unsafe exploratory actions during training. Model-based controllers provide structured behavior, but depend on model fidelity and hand-designed logic. This paper proposes a fading expert-guidance framework that uses a model-based controller to guide DRL training for autonomous overtaking. The expert combines a constrained iterative LQR (CiLQR) planner with PID-based auxiliary controllers, activated when the optimized plan becomes infeasible. The expert action enters the actor objective through a fading term, interpreted as a time-varying soft trust region around the expert policy. This term first biases the policy toward the expert and then vanishes, allowing optimization according to the reinforcement-learning objective. The method is evaluated with PPO, TD3, and SAC. Simulations with bidirectional traffic show improved sample efficiency and final task performance. The safety effects are algorithm-dependent: guidance reduces vehicle collisions for SAC, eliminates boundary collisions for TD3, and leaves the PPO collision profile unchanged. Overall, the results support fading expert guidance as a practical mechanism for transferring model-based driving knowledge to learning-based controllers without restricting the final policy to imitation in an abortable overtaking task with bidirectional traffic.

cs.RO↗
arXiv2026

BR-MPPI: Barrier-Rate Guided MPPI for Enforcing Multiple Inequality Constraints with Learned Signed Distance Fields

Model Predictive Path Integral (MPPI) control provides a sampling-based framework for optimal control, while Control Barrier Functions (CBFs) provide a principled means of enforcing safety constraints. We introduce BR-MPPI, which integrates CBF-like conditions into MPPI's control sampling procedure. CBFs impose inequality constraints that bound the rate of change of barrier functions using a class-K function of the barrier value. We instead impose the CBF condition as an equality constraint using a parametric linear class-K function and augment the system state with its parameter. The parameter's time derivative serves as an additional control input optimized by MPPI. We further design a cost function that promotes parameter values consistent with Nagumo's condition at the safe-set boundary, thereby encouraging safety. The resulting multiple state- and control-dependent equality constraints pose a challenge for random control sampling. We address this through state transformations and control projections inspired by manifold path planning that map sampled controls onto the constraint manifold. We also incorporate learned signed distance fields to represent robot geometry and reduce computation time. Simulations demonstrate improved sample efficiency over vanilla MPPI and higher navigation success rates across five robot models compared with safety-oriented MPPI variants. Hardware experiments on a quadrotor further demonstrate the method's ability to navigate constrained environments near safe-set boundaries.

cs.RO↗
arXiv2026

Tunable Leg Stiffness in a Monopedal Hopper for Energy-Efficient Vertical Hopping Across Varying Ground Profiles

We present the design and implementation of HASTA (Hopper with Adjustable Stiffness for Terrain Adaptation), a vertical hopping robot with real-time tunable leg stiffness, aimed at optimizing energy efficiency across various ground profiles (a pair of ground stiffness and damping conditions). By adjusting leg stiffness, we aim to maximize apex hopping height, a key metric for energy-efficient vertical hopping. We hypothesize that softer legs perform better on soft, damped ground by minimizing penetration and energy loss, while stiffer legs excel on hard, less damped ground by reducing limb deformation and energy dissipation. Through experimental tests and simulations, we find the best leg stiffness within our selection for each combination of ground stiffness and damping, enabling the robot to achieve maximum steady-state hopping height with a constant energy input. These results support our hypothesis that tunable stiffness improves energy-efficient locomotion in controlled experimental conditions. In addition, the simulation provides insights that could aid in the future development of controllers for selecting leg stiffness.

cs.RO↗