Autonomous driving policies are typically trained via open-loop behavior cloning of human demonstrations. However, such policies suffer from covariate shift when deployed in closed loop, leading to compounding errors. W e introduce Rollouts as Demonstrations (RoaD), a simple and efficient method to mitigate covariate shift by leveraging the policy's own closed-loop rollouts as additional training data. During rollout generation, RoaD incorporates expert guidance to bias trajectories toward high-quality behavior, producing informative yet realistic demonstrations for fine-tuning. This approach enables robust closed-loop adaptation with orders of magnitude less data than reinforcement learning, and avoids restrictive assumptions of prior closed-loop supervised fine-tuning (CL-SFT) methods, allowing broader applications domains including end-to-end driving. W e demonstrate the effectiveness of RoaD on WOSAC, a large-scale traffic simulation benchmark, where it performs similar or better than the prior CL-SFT method; and in AlpaSim, a high-fidelity neural reconstruction-based simulator for end-to-end driving, where it improves driving score by 41% and reduces collisions by 54%.

Generation IV (Gen-IV) nuclear power plants are envisioned to replace the current reactor fleet, bringing improvements in performance, safety, reliability, and sustainability. However, large cost investments currently inhibit the deployment of these advanced reactor concepts. Digital twins bridge real-world systems with digital tools to reduce costs, enhance decision-making, and boost operational efficiency. In this work, a digital twin framework is designed to operate the Gen-IV Fluoride-salt-cooled High-temperature Reactor, utilizing data-enhanced methods to optimize operational and maintenance policies while adhering to system constraints. The closed-loop framework integrates surrogate modeling, reinforcement learning, and Bayesian inference to streamline end-to-end communication for online regulation and self-adjustment. Reinforcement learning is used to consider component health and degradation to drive the target power generations, with constraints enforced through a Reference Governor control algorithm that ensures compliance with pump flow rate and temperature limits. These input driving modules benefit from detailed online simulations that are assimilated to measurement data with Bayesian filtering. The digital twin is demonstrated in three case studies: a one-year long-term operational period showcasing maintenance planning capabilities, short-term accuracy refinement with high-frequency measurements, and system shock capturing that demonstrates real-time recalibration capabilities when change in boundary conditions. These demonstrations validate robustness for health-aware and constraint-informed nuclear plant operation, with general applicability to other advanced reactor concepts and complex engineering systems.

Contemporary autoregressive transformers operate in open loop: each hidden state is computed in a single forward pass and never revised, causing errors to propagate uncorrected through the sequence. We identify this open-loop bottleneck as a fundamental architectural limitation underlying well-documented failures in long-range reasoning, factual consistency, and multi-step planning. To address this limitation, we introduce the closed-loop prediction principle, which requires that models iteratively refine latent representations until reaching a self-consistent equilibrium before committing to each token. We instantiate this principle as Equilibrium Transformers (EqT), which augment standard transformer layers with an Equilibrium Refinement Module that minimizes a learned energy function via gradient descent in latent space. The energy function enforces bidirectional prediction consistency, episodic memory coherence, and output confidence, all computed without external supervision. Theoretically, we prove that EqT performs approximate MAP inference in a latent energy-based model, establish linear convergence guarantees, and show that refinement improves predictions precisely on hard instances where one-shot inference is suboptimal. The framework unifies deep equilibrium models, diffusion language models, and test-time training as special cases. Preliminary experiments on the binary parity task demonstrate +3.28% average improvement on challenging sequences, with gains reaching +8.07% where standard transformers approach random performance, validating that the benefit of deliberation scales with task difficulty. Just as attention mechanisms resolved the sequential bottleneck of recurrent networks, we propose that closed-loop equilibrium may resolve the commitment bottleneck of open-loop autoregression, representing a foundational step toward language models.

End-to-end (E2E) autonomous driving models have demonstrated strong performance in open-loop evaluations but often suffer from cascading errors and poor generalization in closed-loop settings. To address this gap, we propose Model-based Policy Adaptation (MPA), a general framework that enhances the robustness and safety of pretrained E2E driving agents during deployment. MPA first generates diverse counterfactual trajectories using a geometry-consistent simulation engine, exposing the agent to scenarios beyond the original dataset. Based on this generated data, MPA trains a diffusion-based policy adapter to refine the base policy's predictions and a multi-step Q value model to evaluate long-term outcomes. At inference time, the adapter proposes multiple trajectory candidates, and the Q value model selects the one with the highest expected utility. Experiments on the nuScenes benchmark using a photorealistic closed-loop simulator demonstrate that MPA significantly improves performance across in-domain, out-of-domain, and safety-critical scenarios. We further investigate how the scale of counterfactual data and inference-time guidance strategies affect overall effectiveness.

Diffusion Policy has shown great performance in robotic manipulation tasks under stochastic perturbations, due to its ability to model multimodal action distributions. Nonetheless, its reliance on a computationally expensive reverse-time diffusion (denoising) process, for action inference, makes it challenging to use for real-time applications where quick decision-making is mandatory. This work studies the possibility of conducting the denoising process only partially before executing an action, allowing the plant to evolve according to its dynamics in parallel to the reverse-time diffusion dynamics ongoing on the computer. In a classical diffusion policy setting, the plant dynamics are usually slow and the two dynamical processes are uncoupled. Here, we investigate theoretical bounds on the stability of closed-loop systems using diffusion policies when the plant dynamics and the denoising dynamics are coupled. The contribution of this work gives a framework for faster imitation learning and a metric that yields if a controller will be stable based on the variance of the demonstrations.

Recent advances in closed-loop planning benchmarks have significantly improved the evaluation of autonomous vehicles. However, existing benchmarks still rely on rule-based reactive agents such as the Intelligent Driver Model (IDM), which lack behavioral diversity and fail to capture realistic human interactions, leading to oversimplified traffic dynamics. To address these limitations, we present nuPlan-R, a new reactive closed-loop planning benchmark that integrates learning-based reactive multi-agent simulation into the nuPlan framework. Our benchmark replaces the rule-based IDM agents with noise-decoupled diffusion-based reactive agents and introduces an interaction-aware agent selection mechanism to ensure both realism and computational efficiency. Furthermore, we extend the benchmark with two additional metrics to enable a more comprehensive assessment of planning performance. Extensive experiments demonstrate that our reactive agent model produces more realistic, diverse, and human-like traffic behaviors, leading to a benchmark environment that better reflects real-world interactive driving. We further reimplement a collection of rule-based, learning-based, and hybrid planning approaches within our nuPlan-R benchmark, providing a clearer reflection of planner performance in complex interactive scenarios and better highlighting the advantages of learning-based planners in handling complex and dynamic scenarios. These results establish nuPlan-R as a new standard for fair, reactive, and realistic closed-loop planning evaluation. We will open-source the code for the new benchmark.

Large Language Models (LLMs) and Vision Language Models (VLMs) have been widely used for embodied symbolic planning. Y et, how to effectively use these models for closed-loop symbolic planning remains largely unexplored. Because they operate as black boxes, LLMs and VLMs can produce unpredictable or costly errors, making their use in high-level robotic planning especially challenging. In this work, we investigate how to use VLMs as closed-loop symbolic planners for robotic applications from a control-theoretic perspective. Concretely, we study how the control horizon and warm-starting impact the performance of VLM symbolic planners. We design and conduct controlled experiments to gain insights that are broadly applicable to utilizing VLMs as closed-loop symbolic planners, and we discuss recommendations that can help improve the performance of VLM symbolic planners. The project website can be found here.

Most recent work in autonomous driving has prioritized benchmark performance and methodological innovation over in-depth analysis of model failures, biases, and shortcut learning. This has led to incremental improvements without a deep understanding of the current failures. While it is straightforward to look at situations where the model fails, it is hard to understand the underlying reason. This motivates us to conduct a systematic study, where inputs to the model are perturbed and the predictions observed. W e introduce PlanT 2.0, a lightweight, object-centric planning transformer designed for autonomous driving research in CARLA. The object-level representation enables controlled analysis, as the input can be easily perturbed (e.g., by changing the location or adding or removing certain objects), in contrast to sensor-based models. T o tackle the scenarios newly introduced by the challenging CARLA Leaderboard 2.0, we introduce multiple upgrades to PlanT, achieving state-of-the-art performance on Longest6 v2, Bench2Drive, and the CARLA validation routes. Our analysis exposes insightful failures, such as a lack of scene understanding caused by low obstacle diversity, rigid expert behaviors leading to exploitable shortcuts, and overfitting to a fixed set of expert trajectories. Based on these findings, we argue for a shift toward data-centric development, with a focus on richer, more robust, and less biased datasets.

Recurrent neural networks (RNNs) trained on neuroscience-inspired tasks offer powerful models of brain computation. However, typical training paradigms rely on open-loop, supervised settings, whereas real-world learning unfolds in closed-loop environments. Here, we develop a mathematical theory describing the learning dynamics of linear RNNs trained in closed-loop contexts. We first demonstrate that two otherwise identical RNNs, trained in either closed- or open-loop modes, follow markedly different learning trajectories. To probe this divergence, we analytically characterize the closed-loop case, revealing distinct stages aligned with the evolution of the training loss. Specifically, we show that the learning dynamics of closed-loop RNNs, in contrast to open-loop ones, are governed by an interplay between two competing objectives: short-term policy improvement and long-term stability of the agent-environment interaction. Finally, we apply our framework to a realistic motor control task, highlighting its broader applicability. Taken together, our results underscore the importance of modeling closed-loop dynamics in a biologically plausible setting.

To move away from open-heart surgery towards safer transcatheter procedures, there is a growing need for improved imaging techniques and robotic solutions to enable simple, accurate tool navigation. Common imaging modalities, such as fluoroscopy and ultrasound, have limitations that can be overcome using cardioscopy, i.e., direct optical visualization inside the beating heart. We present a cardioscope designed as a steerable balloon. As a balloon, it can be collapsed to pass through the vasculature and subsequently inflated inside the heart for visualization and tool delivery through an integrated working channel. Through careful design of balloon wall thickness, a single input, balloon inflation pressure, is used to independently control two outputs, balloon diameter (corresponding to field of view diameter) and balloon bending angle (enabling precise working channel positioning). This balloon technology can be tuned to produce cardioscopes designed for a range of intracardiac tasks. To illustrate this approach, a balloon design is presented for the specific task of aortic leaflet laceration. Image-based closed-loop control of bending angle is also demonstrated as a means of enabling stable orientation control during tool insertion and removal.