Videos are Sample-Efficient Supervisions: Behavior Cloning from Videos via Latent Representations

Humans can efficiently extract knowledge and learn skills from the videos within only a few trials and errors. However, it poses a big challenge to replicate this learning process for autonomous agents, due to the complexity of visual input, the absence of action or reward signals, and the limitations of interaction steps. In this paper, we propose a novel, unsupervised, and sample-efficient framework to achieve imitation learning from videos (ILV), named Behavior Cloning from Videos via Latent Representations (BCV-LR). BCV-LR extracts action-related latent features from high-dimensional video inputs through self-supervised tasks, and then leverages a dynamics-based unsupervised objective to predict latent actions between consecutive frames. The pre-trained latent actions are fine-tuned and efficiently aligned to the real action space online (with collected interactions) for policy behavior cloning. The cloned policy in turn enriches the agent experience for further latent action finetuning, resulting in an iterative policy improvement that is highly sample-efficient. We conduct extensive experiments on a set of challenging visual tasks, including both discrete control and continuous control. BCV-LR enables effective (even expert-level on some tasks) policy performance with only a few interactions, surpassing state-of-the-art ILV baselines and reinforcement learning methods (provided with environmental rewards) in terms of sample efficiency across 24/28 tasks. To the best of our knowledge, this work for the first time demonstrates that videos can support extremely sample-efficient visual policy learning, without the need to access any other expert supervision.

Proprioception Enhances Vision Language Model in Generating Captions and Subtask Segmentations for Robot Task

From the perspective of future developments in robotics, it is crucial to verify whether foundation models trained exclusively on offline data, such as images and language, can understand the robot motion. In particular, since Vision Language Models (VLMs) do not include low-level motion information from robots in their training datasets, video understanding including trajectory information remains a significant challenge. In this study, we assess two capabilities of VLMs through a video captioning task with low-level robot motion information: (1) automatic captioning of robot tasks and (2) segmentation of a series of tasks. Both capabilities are expected to enhance the efficiency of robot imitation learning by linking language and motion and serve as a measure of the foundation model's performance. The proposed method generates multiple "scene" captions using image captions and trajectory data from robot tasks. The full task caption is then generated by summarizing these individual captions. Additionally, the method performs subtask segmentation by comparing the similarity between text embeddings of image captions. In both captioning tasks, the proposed method aims to improve performance by providing the robot's motion data - joint and end-effector states - as input to the VLM. Simulator experiments were conducted to validate the effectiveness of the proposed method.

RoboCade: Gamifying Robot Data Collection

Imitation learning from human demonstrations has become a dominant approach for training autonomous robot policies. However, collecting demonstration datasets is costly: it often requires access to robots and needs sustained effort in a tedious, long process. These factors limit the scale of data available for training policies. We aim to address this scalability challenge by involving a broader audience in a gamified data collection experience that is both accessible and motivating. Specifically, we develop a gamified remote teleoperation platform, RoboCade, to engage general users in collecting data that is beneficial for downstream policy training. To do this, we embed gamification strategies into the design of the system interface and data collection tasks. In the system interface, we include components such as visual feedback, sound effects, goal visualizations, progress bars, leaderboards, and badges. We additionally propose principles for constructing gamified tasks that have overlapping structure with useful downstream target tasks. We instantiate RoboCade on three manipulation tasks -- including spatial arrangement, scanning, and insertion. To illustrate the viability of gamified robot data collection, we collect a demonstration dataset through our platform, and show that co-training robot policies with this data can improve success rate on non-gamified target tasks (+16-56%). Further, we conduct a user study to validate that novice users find the gamified platform significantly more enjoyable than a standard non-gamified platform (+24%). These results highlight the promise of gamified data collection as a scalable, accessible, and engaging method for collecting demonstration data.

LEAD: Minimizing Learner-Expert Asymmetry in End-to-End Driving

Simulators can generate virtually unlimited driving data, yet imitation learning policies in simulation still struggle to achieve robust closed-loop performance. Motivated by this gap, we empirically study how misalignment between privileged expert demonstrations and sensor-based student observations can limit the effectiveness of imitation learning. More precisely, experts have significantly higher visibility (e.g., ignoring occlusions) and far lower uncertainty (e.g., knowing other vehicles' actions), making them difficult to imitate reliably. Furthermore, navigational intent (i.e., the route to follow) is under-specified in student models at test time via only a single target point. We demonstrate that these asymmetries can measurably limit driving performance in CARLA and offer practical interventions to address them. After careful modifications to narrow the gaps between expert and student, our TransFuser v6 (TFv6) student policy achieves a new state of the art on all major publicly available CARLA closed-loop benchmarks, reaching 95 DS on Bench2Drive and more than doubling prior performances on Longest6~v2 and Town13. Additionally, by integrating perception supervision from our dataset into a shared sim-to-real pipeline, we show consistent gains on the NAVSIM and Waymo Vision-Based End-to-End driving benchmarks. Our code, data, and models are publicly available at https://github.com/autonomousvision/lead.

SD2AIL: Adversarial Imitation Learning from Synthetic Demonstrations via Diffusion Models

Adversarial Imitation Learning (AIL) is a dominant framework in imitation learning that infers rewards from expert demonstrations to guide policy optimization. Although providing more expert demonstrations typically leads to improved performance and greater stability, collecting such demonstrations can be challenging in certain scenarios. Inspired by the success of diffusion models in data generation, we propose SD2AIL, which utilizes synthetic demonstrations via diffusion models. We first employ a diffusion model in the discriminator to generate synthetic demonstrations as pseudo-expert data that augment the expert demonstrations. To selectively replay the most valuable demonstrations from the large pool of (pseudo-) expert demonstrations, we further introduce a prioritized expert demonstration replay strategy (PEDR). The experimental results on simulation tasks demonstrate the effectiveness and robustness of our method. In particular, in the Hopper task, our method achieves an average return of 3441, surpassing the state-of-the-art method by 89. Our code will be available at https://github.com/positron-lpc/SD2AIL.

Learning Generalizable Hand-Object Tracking from Synthetic Demonstrations

We present a system for learning generalizable hand-object tracking controllers purely from synthetic data, without requiring any human demonstrations. Our approach makes two key contributions: (1) HOP, a Hand-Object Planner, which can synthesize diverse hand-object trajectories; and (2) HOT, a Hand-Object Tracker that bridges synthetic-to-physical transfer through reinforcement learning and interaction imitation learning, delivering a generalizable controller conditioned on target hand-object states. Our method extends to diverse object shapes and hand morphologies. Through extensive evaluations, we show that our approach enables dexterous hands to track challenging, long-horizon sequences including object re-arrangement and agile in-hand reorientation. These results represent a significant step toward scalable foundation controllers for manipulation that can learn entirely from synthetic data, breaking the data bottleneck that has long constrained progress in dexterous manipulation.

TakeAD: Preference-based Post-optimization for End-to-end Autonomous Driving with Expert Takeover Data

Existing end-to-end autonomous driving methods typically rely on imitation learning (IL) but face a key challenge: the misalignment between open-loop training and closed-loop deployment. This misalignment often triggers driver-initiated takeovers and system disengagements during closed-loop execution. How to leverage those expert takeover data from disengagement scenarios and effectively expand the IL policy's capability presents a valuable yet unexplored challenge. In this paper, we propose TakeAD, a novel preference-based post-optimization framework that fine-tunes the pre-trained IL policy with this disengagement data to enhance the closed-loop driving performance. First, we design an efficient expert takeover data collection pipeline inspired by human takeover mechanisms in real-world autonomous driving systems. Then, this post optimization framework integrates iterative Dataset Aggregation (DAgger) for imitation learning with Direct Preference Optimization (DPO) for preference alignment. The DAgger stage equips the policy with fundamental capabilities to handle disengagement states through direct imitation of expert interventions. Subsequently, the DPO stage refines the policy's behavior to better align with expert preferences in disengagement scenarios. Through multiple iterations, the policy progressively learns recovery strategies for disengagement states, thereby mitigating the open-loop gap. Experiments on the closed-loop Bench2Drive benchmark demonstrate our method's effectiveness compared with pure IL methods, with comprehensive ablations confirming the contribution of each component.

DTCCL: Disengagement-Triggered Contrastive Continual Learning for Autonomous Bus Planners

Autonomous buses run on fixed routes but must operate in open, dynamic urban environments. Disengagement events on these routes are often geographically concentrated and typically arise from planner failures in highly interactive regions. Such policy-level failures are difficult to correct using conventional imitation learning, which easily overfits to sparse disengagement data. To address this issue, this paper presents a Disengagement-Triggered Contrastive Continual Learning (DTCCL) framework that enables autonomous buses to improve planning policies through real-world operation. Each disengagement triggers cloud-based data augmentation that generates positive and negative samples by perturbing surrounding agents while preserving route context. Contrastive learning refines policy representations to better distinguish safe and unsafe behaviors, and continual updates are applied in a cloud-edge loop without human supervision. Experiments on urban bus routes demonstrate that DTCCL improves overall planning performance by 48.6 percent compared with direct retraining, validating its effectiveness for scalable, closed-loop policy improvement in autonomous public transport.

Are All Data Necessary? Efficient Data Pruning for Large-scale Autonomous Driving Dataset via Trajectory Entropy Maximization

Collecting large-scale naturalistic driving data is essential for training robust autonomous driving planners. However, real-world datasets often contain a substantial amount of repetitive and low-value samples, which lead to excessive storage costs and bring limited benefits to policy learning. To address this issue, we propose an information-theoretic data pruning method that effectively reduces the training data volume without compromising model performance. Our approach evaluates the trajectory distribution information entropy of driving data and iteratively selects high-value samples that preserve the statistical characteristics of the original dataset in a model-agnostic manner. From a theoretical perspective, we show that maximizing trajectory entropy effectively constrains the Kullback-Leibler divergence between the pruned subset and the original data distribution, thereby maintaining generalization ability. Comprehensive experiments on the NuPlan benchmark with a large-scale imitation learning framework demonstrate that the proposed method can reduce the dataset size by up to 40% while maintaining closed-loop performance. This work provides a lightweight and theoretically grounded approach for scalable data management and efficient policy learning in autonomous driving systems.

Offline Reinforcement Learning for End-to-End Autonomous Driving

End-to-end (E2E) autonomous driving models that take only camera images as input and directly predict a future trajectory are appealing for their computational efficiency and potential for improved generalization via unified optimization; however, persistent failure modes remain due to reliance on imitation learning (IL). While online reinforcement learning (RL) could mitigate IL-induced issues, the computational burden of neural rendering-based simulation and large E2E networks renders iterative reward and hyperparameter tuning costly. We introduce a camera-only E2E offline RL framework that performs no additional exploration and trains solely on a fixed simulator dataset. Offline RL offers strong data efficiency and rapid experimental iteration, yet is susceptible to instability from overestimation on out-of-distribution (OOD) actions. To address this, we construct pseudo ground-truth trajectories from expert driving logs and use them as a behavior regularization signal, suppressing imitation of unsafe or suboptimal behavior while stabilizing value learning. Training and closed-loop evaluation are conducted in a neural rendering environment learned from the public nuScenes dataset. Empirically, the proposed method achieves substantial improvements in collision rate and route completion compared with IL baselines. Our code will be available at [URL].