Real-world robots need to adapt their behavior beyond the envelope of their pre-trained policy. Policy finetuning or retraining are options, but they risk catastrophic forgetting, degrading the pretrained policy's base performance. To combat this, we introduce CLAE: Closed-Loop Affine Activation Editing, an inference-time framework for steering the behavior of a frozen policy by editing intermediate activations while keeping the base policy weights and downstream action head untouched. CLAE approaches behavior steering as a closed-loop problem whose outputs edit policy activations that adapt online to the robot state, environment, target behavior, and multi-robot context. It trains a sparse autoencoder over frozen-policy activations, selects behavior-relevant latent features via post-hoc probing, and learns a lightweight RL-based steering policy that applies state-dependent affine edits to selected latents during inference. We validate CLAE on a frozen multi-quadrotor navigation policy trained to perform a single task: navigating robots to a set of goal locations while avoiding obstacles. Through extensive simulations and physical tests, we show that while navigating to their goal positions, CLAE can 1. steer individual robot behavior by controlling each robot's velocity profile; 2. coordinate multirobot behavior by preserving a desired formation; and 3. produce entirely new behavior wherein robots are required to reduce their exposure to surveillance cameras in the environment.

翻译：现实世界的机器人需要在其预训练策略的范围之外调整自身行为。策略微调或重新训练是可选方案，但存在灾难性遗忘的风险，会降低预训练策略的基础性能。为解决这一问题，我们提出了CLAE：闭环仿射激活编辑，一种推理阶段的框架，通过编辑中间激活值来引导冻结策略的行为，同时保持基础策略权重和下游动作头不变。CLAE将行为引导视为一个闭环问题，其输出编辑策略激活值，并能根据机器人状态、环境、目标行为和多机器人上下文进行在线自适应。它在冻结策略的激活值上训练稀疏自编码器，通过事后探针选择行为相关的潜在特征，并学习一个轻量级的基于强化学习的引导策略，该策略在推理期间对选定的潜在特征应用依赖于状态的仿射编辑。我们在一个冻结的多四旋翼导航策略上验证了CLAE，该策略训练用于执行单一任务：在避开障碍物的同时将机器人导航至一组目标位置。通过大量仿真和物理实验，我们证明在导航至目标位置的过程中，CLAE能够：1. 通过控制每个机器人的速度曲线来引导单个机器人的行为；2. 通过保持所需队形来协调多机器人行为；3. 产生全新的行为，即要求机器人减少环境中监控摄像头的暴露。