Chenhao Li

Yarden As, Dhruva Tirumala, René Zurbrügg et al. · deepmind

We investigate what specific design choices enable successful online reinforcement learning (RL) on physical robots. Across 100 real-world training runs on three distinct robotic platforms, we systematically ablate algorithmic, systems, and experimental decisions that are typically left implicit in prior work. We find that some widely used defaults can be harmful, while a set of robust, readily adopted design choices within standard RL practice yield stable learning across tasks and hardware. These results provide the first large-sample empirical study of such design choices, enabling practitioners to deploy online RL with lower engineering effort.

Chenhao Li, Marin Vlastelica, Sebastian Blaes et al.

Learning agile skills is one of the main challenges in robotics. To this end, reinforcement learning approaches have achieved impressive results. These methods require explicit task information in terms of a reward function or an expert that can be queried in simulation to provide a target control output, which limits their applicability. In this work, we propose a generative adversarial method for inferring reward functions from partial and potentially physically incompatible demonstrations for successful skill acquirement where reference or expert demonstrations are not easily accessible. Moreover, we show that by using a Wasserstein GAN formulation and transitions from demonstrations with rough and partial information as input, we are able to extract policies that are robust and capable of imitating demonstrated behaviors. Finally, the obtained skills such as a backflip are tested on an agile quadruped robot called Solo 8 and present faithful replication of hand-held human demonstrations.

Chenhao Li, Sebastian Blaes, Pavel Kolev et al.

Learning diverse skills is one of the main challenges in robotics. To this end, imitation learning approaches have achieved impressive results. These methods require explicitly labeled datasets or assume consistent skill execution to enable learning and active control of individual behaviors, which limits their applicability. In this work, we propose a cooperative adversarial method for obtaining single versatile policies with controllable skill sets from unlabeled datasets containing diverse state transition patterns by maximizing their discriminability. Moreover, we show that by utilizing unsupervised skill discovery in the generative adversarial imitation learning framework, novel and useful skills emerge with successful task fulfillment. Finally, the obtained versatile policies are tested on an agile quadruped robot called Solo 8 and present faithful replications of diverse skills encoded in the demonstrations.

Jin Cheng, Marin Vlastelica, Pavel Kolev et al.

Despite many successful applications of data-driven control in robotics, extracting meaningful diverse behaviors remains a challenge. Typically, task performance needs to be compromised in order to achieve diversity. In many scenarios, task requirements are specified as a multitude of reward terms, each requiring a different trade-off. In this work, we take a constrained optimization viewpoint on the quality-diversity trade-off and show that we can obtain diverse policies while imposing constraints on their value functions which are defined through distinct rewards. In line with previous work, further control of the diversity level can be achieved through an attract-repel reward term motivated by the Van der Waals force. We demonstrate the effectiveness of our method on a local navigation task where a quadruped robot needs to reach the target within a finite horizon. Finally, our trained policies transfer well to the real 12-DoF quadruped robot, Solo12, and exhibit diverse agile behaviors with successful obstacle traversal.

Zhihao Cao, Tianxu An, Chenhao Li et al.

Collaborative transport requires robots to infer partner intent through physical interaction while maintaining stable loco-manipulation. This becomes particularly challenging in complex environments, where interaction signals are difficult to capture and model. We present PAINT, a lightweight yet efficient hierarchical learning framework for partner-agonistic intent-aware collaborative legged transport that infers partner intent directly from proprioceptive feedback. PAINT decouples intent understanding from terrain-robust locomotion: A high-level policy infers the partner interaction wrench using an intent estimator and a teacher-student training scheme, while a low-level locomotion backbone ensures robust execution. This enables lightweight deployment without external force-torque sensing or payload tracking. Extensive simulation and real-world experiments demonstrate compliant cooperative transport across diverse terrains, payloads, and partners. Furthermore, we show that PAINT naturally scales to decentralized multi-robot transport and transfers across robot embodiments by swapping the underlying locomotion backbone. Our results suggest that proprioceptive signals in payload-coupled interaction provide a scalable interface for partner-agnostic intent-aware collaborative transport.

Chengcheng Sun, Chenhao Li, Xiang Lin et al.

Graph neural networks (GNNs) aim to learn well-trained representations in a lower-dimension space for downstream tasks while preserving the topological structures. In recent years, attention mechanism, which is brilliant in the fields of natural language processing and computer vision, is introduced to GNNs to adaptively select the discriminative features and automatically filter the noisy information. To the best of our knowledge, due to the fast-paced advances in this domain, a systematic overview of attention-based GNNs is still missing. To fill this gap, this paper aims to provide a comprehensive survey on recent advances in attention-based GNNs. Firstly, we propose a novel two-level taxonomy for attention-based GNNs from the perspective of development history and architectural perspectives. Specifically, the upper level reveals the three developmental stages of attention-based GNNs, including graph recurrent attention networks, graph attention networks, and graph transformers. The lower level focuses on various typical architectures of each stage. Secondly, we review these attention-based methods following the proposed taxonomy in detail and summarize the advantages and disadvantages of various models. A model characteristics table is also provided for a more comprehensive comparison. Thirdly, we share our thoughts on some open issues and future directions of attention-based GNNs. We hope this survey will provide researchers with an up-to-date reference regarding applications of attention-based GNNs. In addition, to cope with the rapid development in this field, we intend to share the relevant latest papers as an open resource at https://github.com/sunxiaobei/awesome-attention-based-gnns.

Gianluca Sabatini, Chenhao Li, Marco Hutter

Proximal Policy Optimization (PPO) has become the de facto standard for training legged robots, thanks to its robustness and scalability in massively parallel simulation environments like IsaacLab. However, its on-policy nature makes it inherently sample-inefficient, preventing its use for continuous adaptation and fine-tuning on real hardware. Soft Actor-Critic (SAC), by contrast, is an off-policy algorithm that can reuse past experience, making it a natural candidate for sim-to-real transfer workflows where the same algorithm can be used both in simulation and for online learning on the real robot. Despite these advantages, SAC has consistently failed to match PPO's empirical performance in massively parallel training settings. This work identifies the root causes of this gap and introduces targeted modifications, covering policy initialization, timeout-aware critic targets, and multi-step return estimation, that enable SAC to train stably at scale. Evaluated across multiple legged robot platforms and diverse locomotion tasks, our approach closes the performance gap with PPO entirely.

Bohao Xu, Yingzhou Lu, Chenhao Li et al.

In drug discovery, predicting the absorption, distribution, metabolism, excretion, and toxicity (ADMET) properties of small-molecule drugs is critical for ensuring safety and efficacy. However, the process of accurately predicting these properties is often resource-intensive and requires extensive experimental data. To address this challenge, we propose SMILES-Mamba, a two-stage model that leverages both unlabeled and labeled data through a combination of self-supervised pretraining and fine-tuning strategies. The model first pre-trains on a large corpus of unlabeled SMILES strings to capture the underlying chemical structure and relationships, before being fine-tuned on smaller, labeled datasets specific to ADMET tasks. Our results demonstrate that SMILES-Mamba exhibits competitive performance across 22 ADMET datasets, achieving the highest score in 14 tasks, highlighting the potential of self-supervised learning in improving molecular property prediction. This approach not only enhances prediction accuracy but also reduces the dependence on large, labeled datasets, offering a promising direction for future research in drug discovery.

Yingzhou Lu, Yaojun Hu, Chenhao Li

Bringing a novel drug from the original idea to market typically requires more than ten years and billions of dollars. To alleviate the heavy burden, a natural idea is to reuse the approved drug to treat new diseases. The process is also known as drug repurposing or drug repositioning. Machine learning methods exhibited huge potential in automating drug repurposing. However, it still encounter some challenges, such as lack of labels and multimodal feature representation. To address these issues, we design DrugCLIP, a cutting-edge contrastive learning method, to learn drug and disease's interaction without negative labels. Additionally, we have curated a drug repurposing dataset based on real-world clinical trial records. Thorough empirical studies are conducted to validate the effectiveness of the proposed DrugCLIP method.

Tao Liu, Zhi Wang, Hui He et al.

The conventional machine learning (ML) and deep learning approaches need to share customers' sensitive information with an external credit bureau to generate a prediction model that opens the door to privacy leakage. This leakage risk makes financial companies face an enormous challenge in their cooperation. Federated learning is a machine learning setting that can protect data privacy, but the high communication cost is often the bottleneck of the federated systems, especially for large neural networks. Limiting the number and size of communications is necessary for the practical training of large neural structures. Gradient sparsification has received increasing attention as a method to reduce communication cost, which only updates significant gradients and accumulates insignificant gradients locally. However, the secure aggregation framework cannot directly use gradient sparsification. This article proposes two sparsification methods to reduce communication cost in federated learning. One is a time-varying hierarchical sparsification method for model parameter update, which solves the problem of maintaining model accuracy after high ratio sparsity. It can significantly reduce the cost of a single communication. The other is to apply the sparsification method to the secure aggregation framework. We sparse the encryption mask matrix to reduce the cost of communication while protecting privacy. Experiments show that under different Non-IID experiment settings, our method can reduce the upload communication cost to about 2.9% to 18.9% of the conventional federated learning algorithm when the sparse rate is 0.01.

Zewei Zhang, Kehan Wen, Michael Xu et al.

Whole-body humanoid locomotion is challenging due to high-dimensional control, morphological instability, and the need for real-time adaptation to various terrains using onboard perception. Directly applying reinforcement learning (RL) with reward shaping to humanoid locomotion often leads to lower-body-dominated behaviors, whereas imitation-based RL can learn more coordinated whole-body skills but is typically limited to replaying reference motions without a mechanism to adapt them online from perception for terrain-aware locomotion. To address this gap, we propose a whole-body humanoid locomotion framework that combines skills learned from reference motions with terrain-aware adaptation. We first train a diffusion model on retargeted human motions for real-time prediction of terrain-aware reference motions. Concurrently, we train a whole-body reference tracker with RL using this motion data. To improve robustness under imperfectly generated references, we further fine-tune the tracker with a frozen motion generator in a closed-loop setting. The resulting system supports directional goal-reaching control with terrain-aware whole-body adaptation, and can be deployed on a Unitree G1 humanoid robot with onboard perception and computation. The hardware experiments demonstrate successful traversal over boxes, hurdles, stairs, and mixed terrain combinations. Quantitative results further show the benefits of incorporating online motion generation and fine-tuning the motion tracker for improved generalization and robustness.

Hehui Zheng, Bhavya Sukhija, Chenhao Li et al.

Soft robots offer unmatched adaptability and safety in unstructured environments, yet their compliant, high-dimensional, and nonlinear dynamics make modeling for control notoriously difficult. Existing data-driven approaches often fail to generalize, constrained by narrowly focused task demonstrations or inefficient random exploration. We introduce SoftAE, an uncertainty-aware active exploration framework that autonomously learns task-agnostic and generalizable dynamics models of soft robotic systems. SoftAE employs probabilistic ensemble models to estimate epistemic uncertainty and actively guides exploration toward underrepresented regions of the state-action space, achieving efficient coverage of diverse behaviors without task-specific supervision. We evaluate SoftAE on three simulated soft robotic platforms -- a continuum arm, an articulated fish in fluid, and a musculoskeletal leg with hybrid actuation -- and on a pneumatically actuated continuum soft arm in the real world. Compared with random exploration and task-specific model-based reinforcement learning, SoftAE produces more accurate dynamics models, enables superior zero-shot control on unseen tasks, and maintains robustness under sensing noise, actuation delays, and nonlinear material effects. These results demonstrate that uncertainty-driven active exploration can yield scalable, reusable dynamics models across diverse soft robotic morphologies, representing a step toward more autonomous, adaptable, and data-efficient control in compliant robots.

Chenhao Li, Taishi Ono, Takeshi Uemori et al.

Multi-view inverse rendering is the problem of estimating the scene parameters such as shapes, materials, or illuminations from a sequence of images captured under different viewpoints. Many approaches, however, assume single light bounce and thus fail to recover challenging scenarios like inter-reflections. On the other hand, simply extending those methods to consider multi-bounced light requires more assumptions to alleviate the ambiguity. To address this problem, we propose Neural Incident Stokes Fields (NeISF), a multi-view inverse rendering framework that reduces ambiguities using polarization cues. The primary motivation for using polarization cues is that it is the accumulation of multi-bounced light, providing rich information about geometry and material. Based on this knowledge, the proposed incident Stokes field efficiently models the accumulated polarization effect with the aid of an original physically-based differentiable polarimetric renderer. Lastly, experimental results show that our method outperforms the existing works in synthetic and real scenarios.

Yaru Niu, Yunzhe Zhang, Mingyang Yu et al.

Quadrupedal robots have demonstrated impressive locomotion capabilities in complex environments, but equipping them with autonomous versatile manipulation skills in a scalable way remains a significant challenge. In this work, we introduce a cross-embodiment imitation learning system for quadrupedal manipulation, leveraging data collected from both humans and LocoMan, a quadruped equipped with multiple manipulation modes. Specifically, we develop a teleoperation and data collection pipeline, which unifies and modularizes the observation and action spaces of the human and the robot. To effectively leverage the collected data, we propose an efficient modularized architecture that supports co-training and pretraining on structured modality-aligned data across different embodiments. Additionally, we construct the first manipulation dataset for the LocoMan robot, covering various household tasks in both unimanual and bimanual modes, supplemented by a corresponding human dataset. We validate our system on six real-world manipulation tasks, where it achieves an average success rate improvement of 41.9% overall and 79.7% under out-of-distribution (OOD) settings compared to the baseline. Pretraining with human data contributes a 38.6% success rate improvement overall and 82.7% under OOD settings, enabling consistently better performance with only half the amount of robot data. Our code, hardware, and data are open-sourced at: https://human2bots.github.io.

Yingzhou Lu, Minjie Shen, Ling Yue et al.

The surge in high-throughput omics data has reshaped the landscape of biological research, underlining the need for powerful, user-friendly data analysis and interpretation tools. This paper presents GenoCraft, a web-based comprehensive software solution designed to handle the entire pipeline of omics data processing. GenoCraft offers a unified platform featuring advanced bioinformatics tools, covering all aspects of omics data analysis. It encompasses a range of functionalities, such as normalization, quality control, differential analysis, network analysis, pathway analysis, and diverse visualization techniques. This software makes state-of-the-art omics data analysis more accessible to a wider range of users. With GenoCraft, researchers and data scientists have access to an array of cutting-edge bioinformatics tools under a user-friendly interface, making it a valuable resource for managing and analyzing large-scale omics data. The API with an interactive web interface is publicly available at https://genocraft.stanford. edu/. We also release all the codes in https://github.com/futianfan/GenoCraft.

Luan Zhang, Dandan Song, Zhijing Wu et al.

Tool-integrated reasoning (TIR) enables large language models (LLMs) to enhance their capabilities by interacting with external tools, such as code interpreters (CI). Most recent studies focus on exploring various methods to equip LLMs with the ability to use tools. However, how to further boost the reasoning ability of already tool-capable LLMs at inference time remains underexplored. Improving reasoning at inference time requires no additional training and can help LLMs better leverage tools to solve problems. We observe that, during tool-capable LLM inference, both the number and the proportion of erroneous tool calls are negatively correlated with answer correctness. Moreover, erroneous tool calls are typically resolved successfully within a few subsequent turns. If not, LLMs often struggle to resolve such errors even with many additional turns. Building on the above observations, we propose PruneTIR, a rather effective yet efficient framework that enhances the tool-integrated reasoning at inference time. During LLM inference, PruneTIR prunes trajectories, resamples tool calls, and suspends tool usage through three components: Success-Triggered Pruning, Stuck-Triggered Pruning and Resampling, and Retry-Triggered Tool Suspension. These three components enable PruneTIR to mitigate the negative impact of erroneous tool calls and prevent LLMs from getting stuck in repeated failed resolution attempts, thereby improving overall LLM performance. Extensive experimental results demonstrate the effectiveness of PruneTIR, which significantly improves Pass@1 and efficiency while reducing the working context length for tool-capable LLMs.

Mohamad H. Danesh, Chenhao Li, Amin Abyaneh et al.

World models promise a paradigm shift in robotics, where an agent learns the underlying physics of its environment once to enable efficient planning and behavior learning. However, current world models are often hardware-locked specialists: a model trained on a Boston Dynamics Spot robot fails catastrophically on a Unitree Go1 due to the mismatch in kinematic and dynamic properties, as the model overfits to specific embodiment constraints rather than capturing the universal locomotion dynamics. Consequently, a slight change in actuator dynamics or limb length necessitates training a new model from scratch. In this work, we take a step towards a framework for training a generalizable Quadrupedal World Model (QWM) that disentangles environmental dynamics from robot morphology. We address the limitations of implicit system identification, where treating static physical properties (like mass or limb length) as latent variables to be inferred from motion history creates an adaptation lag that can compromise zero-shot safety and efficiency. Instead, we explicitly condition the generative dynamics on the robot's engineering specifications. By integrating a physical morphology encoder and a reward normalizer, we enable the model to serve as a neural simulator capable of generalizing across morphologies. This capability unlocks zero-shot control across a range of embodiments. We introduce, for the first time, a world model that enables zero-shot generalization to new morphologies for locomotion. While we carefully study the limitations of our method, QWM operates as a distribution-bounded interpolator within the quadrupedal morphology family rather than a universal physics engine, this work represents a significant step toward morphology-conditioned world models for legged locomotion.

Huipeng Ma, Luan Zhang, Dandan Song et al.

In multi-hop reasoning, multi-round retrieval-augmented generation (RAG) methods typically rely on LLM-generated content as the retrieval query. However, these approaches are inherently vulnerable to knowledge overshadowing - a phenomenon where critical information is overshadowed during generation. As a result, the LLM-generated content may be incomplete or inaccurate, leading to irrelevant retrieval and causing error accumulation during the iteration process. To address this challenge, we propose ActiShade, which detects and activates overshadowed knowledge to guide large language models (LLMs) in multi-hop reasoning. Specifically, ActiShade iteratively detects the overshadowed keyphrase in the given query, retrieves documents relevant to both the query and the overshadowed keyphrase, and generates a new query based on the retrieved documents to guide the next-round iteration. By supplementing the overshadowed knowledge during the formulation of next-round queries while minimizing the introduction of irrelevant noise, ActiShade reduces the error accumulation caused by knowledge overshadowing. Extensive experiments show that ActiShade outperforms existing methods across multiple datasets and LLMs.

Chenhao Li, Trung Thanh Ngo, Hajime Nagahara

In this work, we propose a novel learning-based method to jointly estimate the shape and subsurface scattering (SSS) parameters of translucent objects by utilizing polarization cues. Although polarization cues have been used in various applications, such as shape from polarization (SfP), BRDF estimation, and reflection removal, their application in SSS estimation has not yet been explored. Our observations indicate that the SSS affects not only the light intensity but also the polarization signal. Hence, the polarization signal can provide additional cues for SSS estimation. We also introduce the first large-scale synthetic dataset of polarized translucent objects for training our model. Our method outperforms several baselines from the SfP and inverse rendering realms on both synthetic and real data, as demonstrated by qualitative and quantitative results.

Edoardo Debenedetti, Javier Rando, Daniel Paleka et al.

Large language model systems face important security risks from maliciously crafted messages that aim to overwrite the system's original instructions or leak private data. To study this problem, we organized a capture-the-flag competition at IEEE SaTML 2024, where the flag is a secret string in the LLM system prompt. The competition was organized in two phases. In the first phase, teams developed defenses to prevent the model from leaking the secret. During the second phase, teams were challenged to extract the secrets hidden for defenses proposed by the other teams. This report summarizes the main insights from the competition. Notably, we found that all defenses were bypassed at least once, highlighting the difficulty of designing a successful defense and the necessity for additional research to protect LLM systems. To foster future research in this direction, we compiled a dataset with over 137k multi-turn attack chats and open-sourced the platform.

Yun Tian, Guili Wang, Jian Bi et al.

Accurate prediction of lithium-ion battery remaining useful life (RUL) is essential for reliable health monitoring and data-driven analysis of battery degradation. However, the robustness and generalization capabilities of existing RUL prediction models are significantly challenged by complex operating conditions and limited data availability. To address these limitations, this study proposes a hybrid deep learning model, CDFormer, which integrates convolutional neural networks, deep residual shrinkage networks, and Transformer encoders extract multiscale temporal features from battery measurement signals, including voltage, current, and capacity. This architecture enables the joint modeling of local and global degradation dynamics, effectively improving the accuracy of RUL prediction.To enhance predictive reliability, a composite temporal data augmentation strategy is proposed, incorporating Gaussian noise, time warping, and time resampling, explicitly accounting for measurement noise and variability. CDFormer is evaluated on two real-world datasets, with experimental results demonstrating its consistent superiority over conventional recurrent neural network-based and Transformer-based baselines across key metrics. By improving the reliability and predictive performance of RUL prediction from measurement data, CDFormer provides accurate and reliable forecasts, supporting effective battery health monitoring and data-driven maintenance strategies.

Chenhao Li, Elijah Stanger-Jones, Steve Heim et al.

Motion trajectories offer reliable references for physics-based motion learning but suffer from sparsity, particularly in regions that lack sufficient data coverage. To address this challenge, we introduce a self-supervised, structured representation and generation method that extracts spatial-temporal relationships in periodic or quasi-periodic motions. The motion dynamics in a continuously parameterized latent space enable our method to enhance the interpolation and generalization capabilities of motion learning algorithms. The motion learning controller, informed by the motion parameterization, operates online tracking of a wide range of motions, including targets unseen during training. With a fallback mechanism, the controller dynamically adapts its tracking strategy and automatically resorts to safe action execution when a potentially risky target is proposed. By leveraging the identified spatial-temporal structure, our work opens new possibilities for future advancements in general motion representation and learning algorithms.

Chenhao Li, Andreas Krause, Marco Hutter

Learning robust and generalizable world models is crucial for enabling efficient and scalable robotic control in real-world environments. In this work, we introduce a novel framework for learning world models that accurately capture complex, partially observable, and stochastic dynamics. The proposed method employs a dual-autoregressive mechanism and self-supervised training to achieve reliable long-horizon predictions without relying on domain-specific inductive biases, ensuring adaptability across diverse robotic tasks. We further propose a policy optimization framework that leverages world models for efficient training in imagined environments and seamless deployment in real-world systems. This work advances model-based reinforcement learning by addressing the challenges of long-horizon prediction, error accumulation, and sim-to-real transfer. By providing a scalable and robust framework, the introduced methods pave the way for adaptive and efficient robotic systems in real-world applications.

Chenhao Li, Andreas Krause, Marco Hutter

Reinforcement Learning (RL) has demonstrated impressive capabilities in robotic control but remains challenging due to high sample complexity, safety concerns, and the sim-to-real gap. While offline RL eliminates the need for risky real-world exploration by learning from pre-collected data, it suffers from distributional shift, limiting policy generalization. Model-Based RL (MBRL) addresses this by leveraging predictive models for synthetic rollouts, yet existing approaches often lack robust uncertainty estimation, leading to compounding errors in offline settings. We introduce Offline Robotic World Model (RWM-O), a model-based approach that explicitly estimates epistemic uncertainty to improve policy learning without reliance on a physics simulator. By integrating these uncertainty estimates into policy optimization, our approach penalizes unreliable transitions, reducing overfitting to model errors and enhancing stability. Experimental results show that RWM-O improves generalization and safety, enabling policy learning purely from real-world data and advancing scalable, data-efficient RL for robotics.

Mert Albaba, Chenhao Li, Markos Diomataris et al.

Acquiring physically plausible motor skills across diverse and unconventional morphologies-including humanoid robots, quadrupeds, and animals-is essential for advancing character simulation and robotics. Traditional methods, such as reinforcement learning (RL) are task- and body-specific, require extensive reward function engineering, and do not generalize well. Imitation learning offers an alternative but relies heavily on high-quality expert demonstrations, which are difficult to obtain for non-human morphologies. Video diffusion models, on the other hand, are capable of generating realistic videos of various morphologies, from humans to ants. Leveraging this capability, we propose a data-independent approach for skill acquisition that learns 3D motor skills from 2D-generated videos, with generalization capability to unconventional and non-human forms. Specifically, we guide the imitation learning process by leveraging vision transformers for video-based comparisons by calculating pair-wise distance between video embeddings. Along with video-encoding distance, we also use a computed similarity between segmented video frames as a guidance reward. We validate our method on locomotion tasks involving unique body configurations. In humanoid robot locomotion tasks, we demonstrate that 'No-data Imitation Learning' (NIL) outperforms baselines trained on 3D motion-capture data. Our results highlight the potential of leveraging generative video models for physically plausible skill learning with diverse morphologies, effectively replacing data collection with data generation for imitation learning.

Kaixi Bao, Chenhao Li, Yarden As et al.

Agents trained via reinforcement learning (RL) often struggle to perform well on tasks that differ from those encountered during training. This limitation presents a challenge to the broader deployment of RL in diverse and dynamic task settings. In this work, we introduce memory augmentation, a memory-based RL approach to improve task generalization. Our approach leverages task-structured augmentations to simulate plausible out-of-distribution scenarios and incorporates memory mechanisms to enable context-aware policy adaptation. Trained on a predefined set of tasks, our policy demonstrates the ability to generalize to unseen tasks through memory augmentation without requiring additional interactions with the environment. Through extensive simulation experiments and real-world hardware evaluations on legged locomotion tasks, we demonstrate that our approach achieves zero-shot generalization to unseen tasks while maintaining robust in-distribution performance and high sample efficiency.

Chenhao Li, Marco Hutter, Andreas Krause

This survey provides a comparative analysis of feature-based and GAN-based approaches to learning from demonstrations, with a focus on the structure of reward functions and their implications for policy learning. Feature-based methods offer dense, interpretable rewards that excel at high-fidelity motion imitation, yet often require sophisticated representations of references and struggle with generalization in unstructured settings. GAN-based methods, in contrast, use implicit, distributional supervision that enables scalability and adaptation flexibility, but are prone to training instability and coarse reward signals. Recent advancements in both paradigms converge on the importance of structured motion representations, which enable smoother transitions, controllable synthesis, and improved task integration. We argue that the dichotomy between feature-based and GAN-based methods is increasingly nuanced: rather than one paradigm dominating the other, the choice should be guided by task-specific priorities such as fidelity, diversity, interpretability, and adaptability. This work outlines the algorithmic trade-offs and design considerations that underlie method selection, offering a framework for principled decision-making in learning from demonstrations.

Chenhao Li, Taishi Ono, Takeshi Uemori et al.

Recent inverse rendering methods have greatly improved shape, material, and illumination reconstruction by utilizing polarization cues. However, existing methods only support dielectrics, ignoring conductors that are found everywhere in life. Since conductors and dielectrics have different reflection properties, using previous conductor methods will lead to obvious errors. In addition, conductors are glossy, which may cause strong specular reflection and is hard to reconstruct. To solve the above issues, we propose NeISF++, an inverse rendering pipeline that supports conductors and dielectrics. The key ingredient for our proposal is a general pBRDF that describes both conductors and dielectrics. As for the strong specular reflection problem, we propose a novel geometry initialization method using DoLP images. This physical cue is invariant to intensities and thus robust to strong specular reflections. Experimental results on our synthetic and real datasets show that our method surpasses the existing polarized inverse rendering methods for geometry and material decomposition as well as downstream tasks like relighting.

Klemens Iten, Bruce Lee, Chenhao Li et al.

Learning-based control methods typically assume stationary system dynamics, an assumption often violated in real-world systems due to drift, wear, or changing operating conditions. We study reinforcement learning for control under time-varying dynamics. We consider a continual model-based reinforcement learning setting in which an agent repeatedly learns and controls a dynamical system whose transition dynamics evolve across episodes. We analyze the problem using Gaussian process dynamics models under frequentist variation-budget assumptions. Our analysis shows that persistent non-stationarity requires explicitly limiting the influence of outdated data to maintain calibrated uncertainty and meaningful dynamic regret guarantees. Motivated by these insights, we propose a practical optimistic model-based reinforcement learning algorithm with adaptive data buffer mechanisms and demonstrate improved performance on continuous control benchmarks with non-stationary dynamics.

Chenhao Li, Taishi Ono, Takeshi Uemori et al.

We show that, with polarization cues, a lightweight model trained on a small dataset can outperform RGB-only vision foundation models (VFMs) in single-shot object-level surface normal estimation. Shape from polarization (SfP) has long been studied due to the strong physical relationship between polarization and surface geometry. Meanwhile, driven by scaling laws, RGB-only VFMs trained on large datasets have recently achieved impressive performance and surpassed existing SfP methods. This situation raises questions about the necessity of polarization cues, which require specialized hardware and have limited training data. We argue that the weaker performance of prior SfP methods does not come from the polarization modality itself, but from domain gaps. These domain gaps mainly arise from two sources. First, existing synthetic datasets use limited and unrealistic 3D objects, with simple geometry and random texture maps that do not match the underlying shapes. Second, real-world polarization signals are often affected by sensor noise, which is not well modeled during training. To address the first issue, we render a high-quality polarization dataset using 1,954 3D-scanned real-world objects. We further incorporate pretrained DINOv3 priors to improve generalization to unseen objects. To address the second issue, we introduce polarization sensor-aware data augmentation that better reflects real-world conditions. With only 40K training scenes, our method significantly outperforms both state-of-the-art SfP approaches and RGB-only VFMs. Extensive experiments show that polarization cues enable a 33x reduction in training data or an 8x reduction in model parameters, while still achieving better performance than RGB-only counterparts.

Chenhao Li, Trung Thanh Ngo, Hajime Nagahara

In this work, we propose an inverse rendering model that estimates 3D shape, spatially-varying reflectance, homogeneous subsurface scattering parameters, and an environment illumination jointly from only a pair of captured images of a translucent object. In order to solve the ambiguity problem of inverse rendering, we use a physically-based renderer and a neural renderer for scene reconstruction and material editing. Because two renderers are differentiable, we can compute a reconstruction loss to assist parameter estimation. To enhance the supervision of the proposed neural renderer, we also propose an augmented loss. In addition, we use a flash and no-flash image pair as the input. To supervise the training, we constructed a large-scale synthetic dataset of translucent objects, which consists of 117K scenes. Qualitative and quantitative results on both synthetic and real-world datasets demonstrated the effectiveness of the proposed model.