Zhenshan Bing

Xiangtong Yao, Hongkuan Zhou, Oier Mees et al. · cmu

Language-conditioned robot manipulation is an emerging field aimed at enabling seamless communication and cooperation between humans and robotic agents by teaching robots to comprehend and execute instructions conveyed in natural language. This interdisciplinary area integrates scene understanding, language processing, and policy learning to bridge the gap between human instructions and robot actions. In this comprehensive survey, we systematically explore recent advancements in language-conditioned robot manipulation. We categorize existing methods based on the primary ways language is integrated into the robot system, namely language for state evaluation, language as a policy condition, language for cognitive planning and reasoning, and language in unified vision-language-action models. Specifically, we further analyze state-of-the-art techniques from five axes of action granularity, data and supervision regimes, system cost and latency, environments and evaluations, and cross-modal task specification. Additionally, we highlight the key debates in the field. Finally, we discuss open challenges and future research directions, focusing on potentially enhancing generalization capabilities and addressing safety issues in language-conditioned robot manipulators.

Shengkai Liu, Yaofeng Cheng, Fusheng Zha et al.

Pruning at initialization (PaI) reduces training costs by removing weights before training, which becomes increasingly crucial with the growing network size. However, current PaI methods still have a large accuracy gap with iterative pruning, especially at high sparsity levels. This raises an intriguing question: can we get inspiration from iterative pruning to improve the PaI performance? In the lottery ticket hypothesis, the iterative rewind pruning (IRP) finds subnetworks retroactively by rewinding the parameter to the original initialization in every pruning iteration, which means all the subnetworks are based on the initial state. Here, we hypothesise the surviving subnetworks are more important and bridge the initial feature and their surviving score as the PaI criterion. We employ an end-to-end neural network (\textbf{AutoS}parse) to learn this correlation, input the model's initial features, output their score and then prune the lowest score parameters before training. To validate the accuracy and generalization of our method, we performed PaI across various models. Results show that our approach outperforms existing methods in high-sparsity settings. Notably, as the underlying logic of model pruning is consistent in different models, only one-time IRP on one model is needed (e.g., once IRP on ResNet-18/CIFAR-10, AutoS can be generalized to VGG-16/CIFAR-10, ResNet-18/TinyImageNet, et al.). As the first neural network-based PaI method, we conduct extensive experiments to validate the factors influencing this approach. These results reveal the learning tendencies of neural networks and provide new insights into our understanding and research of PaI from a practical perspective. Our code is available at: https://github.com/ChengYaofeng/AutoSparse.git.

Xiangtong Yao, Zhenshan Bing, Genghang Zhuang et al.

Meta-reinforcement learning (meta-RL) is a promising approach that enables the agent to learn new tasks quickly. However, most meta-RL algorithms show poor generalization in multi-task scenarios due to the insufficient task information provided only by rewards. Language-conditioned meta-RL improves the generalization capability by matching language instructions with the agent's behaviors. While both behaviors and language instructions have symmetry, which can speed up human learning of new knowledge. Thus, combining symmetry and language instructions into meta-RL can help improve the algorithm's generalization and learning efficiency. We propose a dual-MDP meta-reinforcement learning method that enables learning new tasks efficiently with symmetrical behaviors and language instructions. We evaluate our method in multiple challenging manipulation tasks, and experimental results show that our method can greatly improve the generalization and learning efficiency of meta-reinforcement learning. Videos are available at https://tumi6robot.wixsite.com/symmetry/.

Mingyang Wang, Zhenshan Bing, Xiangtong Yao et al.

Meta-reinforcement learning enables artificial agents to learn from related training tasks and adapt to new tasks efficiently with minimal interaction data. However, most existing research is still limited to narrow task distributions that are parametric and stationary, and does not consider out-of-distribution tasks during the evaluation, thus, restricting its application. In this paper, we propose MoSS, a context-based Meta-reinforcement learning algorithm based on Self-Supervised task representation learning to address this challenge. We extend meta-RL to broad non-parametric task distributions which have never been explored before, and also achieve state-of-the-art results in non-stationary and out-of-distribution tasks. Specifically, MoSS consists of a task inference module and a policy module. We utilize the Gaussian mixture model for task representation to imitate the parametric and non-parametric task variations. Additionally, our online adaptation strategy enables the agent to react at the first sight of a task change, thus being applicable in non-stationary tasks. MoSS also exhibits strong generalization robustness in out-of-distributions tasks which benefits from the reliable and robust task representation. The policy is built on top of an off-policy RL algorithm and the entire network is trained completely off-policy to ensure high sample efficiency. On MuJoCo and Meta-World benchmarks, MoSS outperforms prior works in terms of asymptotic performance, sample efficiency (3-50x faster), adaptation efficiency, and generalization robustness on broad and diverse task distributions.

Hongkuan Zhou, Wei Cao, Aifen Sui et al.

End-to-end autonomous driving, where the entire driving pipeline is replaced with a single neural network, has recently gained research attention because of its simpler structure and faster inference time. Despite this appealing approach largely reducing the complexity in the driving pipeline, it also leads to safety issues because the trained policy is not always compliant with the traffic rules. In this paper, we proposed P-CSG, a penalty-based imitation learning approach with contrastive-based cross semantics generation sensor fusion technologies to increase the overall performance of end-to-end autonomous driving. In this method, we introduce three penalties - red light, stop sign, and curvature speed penalty to make the agent more sensitive to traffic rules. The proposed cross semantics generation helps to align the shared information of different input modalities. We assessed our model's performance using the CARLA Leaderboard - Town 05 Long Benchmark and Longest6 Benchmark, achieving 8.5% and 2.0% driving score improvement compared to the baselines. Furthermore, we conducted robustness evaluations against adversarial attacks like FGSM and Dot attacks, revealing a substantial increase in robustness compared to other baseline models. More detailed information can be found at https://hk-zh.github.io/p-csg-plus.

Ju Dong, Liding Zhang, Lei Zhang et al.

Generative policies based on diffusion and flow matching achieve strong performance in robotic manipulation by modeling multi-modal human demonstrations. However, their reliance on iterative Ordinary Differential Equation (ODE) integration introduces substantial latency, limiting high-frequency closed-loop control. Recent single-step acceleration methods alleviate this overhead but often exhibit distributional collapse, producing averaged trajectories that fail to execute coherent manipulation strategies. We propose a framework that distills a Conditional Flow Matching (CFM) expert into a fast single-step student via Implicit Maximum Likelihood Estimation (IMLE). A bi-directional Chamfer distance provides a set-level objective that promotes both mode coverage and fidelity, enabling preservation of the teacher multi-modal action distribution in a single forward pass. A unified perception encoder further integrates multi-view RGB, depth, point clouds, and proprioception into a geometry-aware representation. The resulting high-frequency control supports real-time receding-horizon re-planning and improved robustness under dynamic disturbances.

Lei Zhang, Diwen Zheng, Kaixin Bai et al.

Dexterous grasp generation aims to produce grasp poses that align with task requirements and human interpretable grasp semantics. However, achieving semantically controllable dexterous grasp synthesis remains highly challenging due to the lack of unified modeling of multiple semantic dimensions, including grasp taxonomy, contact semantics, and functional affordance. To address these limitations, we present OmniDexVLG, a multimodal, semantics aware grasp generation framework capable of producing structurally diverse and semantically coherent dexterous grasps under joint language and visual guidance. Our approach begins with OmniDexDataGen, a semantic rich dexterous grasp dataset generation pipeline that integrates grasp taxonomy guided configuration sampling, functional affordance contact point sampling, taxonomy aware differential force closure grasp sampling, and physics based optimization and validation, enabling systematic coverage of diverse grasp types. We further introduce OmniDexReasoner, a multimodal grasp type semantic reasoning module that leverages multi agent collaboration, retrieval augmented generation, and chain of thought reasoning to infer grasp related semantics and generate high quality annotations that align language instructions with task specific grasp intent. Building upon these components, we develop a unified Vision Language Grasping generation model that explicitly incorporates grasp taxonomy, contact structure, and functional affordance semantics, enabling fine grained control over grasp synthesis from natural language instructions. Extensive experiments in simulation and real world object grasping and ablation studies demonstrate that our method substantially outperforms state of the art approaches in terms of grasp diversity, contact semantic diversity, functional affordance diversity, and semantic consistency.

Kejia Chen, Zheng Shen, Yue Zhang et al.

Large Language Models (LLMs) have gained popularity in task planning for long-horizon manipulation tasks. To enhance the validity of LLM-generated plans, visual demonstrations and online videos have been widely employed to guide the planning process. However, for manipulation tasks involving subtle movements but rich contact interactions, visual perception alone may be insufficient for the LLM to fully interpret the demonstration. Additionally, visual data provides limited information on force-related parameters and conditions, which are crucial for effective execution on real robots. In this paper, we introduce an in-context learning framework that incorporates tactile and force-torque information from human demonstrations to enhance LLMs' ability to generate plans for new task scenarios. We propose a bootstrapped reasoning pipeline that sequentially integrates each modality into a comprehensive task plan. This task plan is then used as a reference for planning in new task configurations. Real-world experiments on two different sequential manipulation tasks demonstrate the effectiveness of our framework in improving LLMs' understanding of multi-modal demonstrations and enhancing the overall planning performance.

Haichuan Li, Liguo Zhou, Zhenshan Bing et al.

Several autonomous driving strategies have been applied to autonomous vehicles, especially in the collision avoidance area. The purpose of collision avoidance is achieved by adjusting the trajectory of autonomous vehicles (AV) to avoid intersection or overlap with the trajectory of surrounding vehicles. A large number of sophisticated vision algorithms have been designed for target inspection, classification, and other tasks, such as ResNet, YOLO, etc., which have achieved excellent performance in vision tasks because of their ability to accurately and quickly capture regional features. However, due to the variability of different tasks, the above models achieve good performance in capturing small regions but are still insufficient in correlating the regional features of the input image with each other. In this paper, we aim to solve this problem and develop an algorithm that takes into account the advantages of CNN in capturing regional features while establishing feature correlation between regions using variants of attention. Finally, our model achieves better performance in the test set of L5Kit compared to the other vision models. The average number of collisions is 19.4 per 10000 frames of driving distance, which greatly improves the success rate of collision avoidance.

Ziou Hu, Xiangtong Yao, Yuan Meng et al.

Diffusion policies excel at visuomotor control but often fail catastrophically under severe out-of-distribution (OOD) disturbances, such as unexpected object displacements or visual corruptions. To address this vulnerability, we introduce the Dream Diffusion Policy (DDP), a framework that deeply integrates a diffusion world model into the policy's training objective via a shared 3D visual encoder. This co-optimization endows the policy with robust state-prediction capabilities. When encountering sudden OOD anomalies during inference, DDP detects the real-imagination discrepancy and actively abandons the corrupted visual stream. Instead, it relies on its internal "imagination" (autoregressively forecasted latent dynamics) to safely bypass the disruption, generating imagined trajectories before smoothly realigning with physical reality. Extensive evaluations demonstrate DDP's exceptional resilience. Notably, DDP achieves a 73.8% OOD success rate on MetaWorld (vs. 23.9% without predictive imagination) and an 83.3% success rate under severe real-world spatial shifts (vs. 3.3% without predictive imagination). Furthermore, as a stress test, DDP maintains a 76.7% real-world success rate even when relying entirely on open-loop imagination post-initialization.

Xukun Li, Yu Sun, Lei Zhang et al.

Bimanual dexterous manipulation relies on integrating multimodal inputs to perform complex real-world tasks. To address the challenges of effectively combining these modalities, we propose DECO, a decoupled multimodal diffusion transformer that disentangles vision, proprioception, and tactile signals through specialized conditioning pathways, enabling structured and controllable integration of multimodal inputs, with a lightweight adapter for parameter-efficient injection of additional signals. Alongside DECO, we release DECO-50 dataset for bimanual dexterous manipulation with tactile sensing, consisting of 50 hours of data and over 5M frames, collected via teleoperation on real dual-arm robots. We train DECO on DECO-50 and conduct extensive real-world evaluation with over 2,000 robot rollouts. Experimental results show that DECO achieves the best performance across all tasks, with a 72.25% average success rate and a 21% improvement over the baseline. Moreover, the tactile adapter brings an additional 10.25% average success rate across all tasks and a 20% gain on complex contact-rich tasks while tuning less than 10% of the model parameters.

Lei Liu, Haonan Zhang, Huahang Xu et al.

Spinning flexible objects, exemplified by traditional Chinese handkerchief performances, demands periodic steady-state motions under nonlinear dynamics with frictional contacts and boundary constraints. To address these challenges, we first design an intuitive dexterous wrist based on a parallel anti-parallelogram tendon-driven structure, which achieves 90 degrees omnidirectional rotation with low inertia and decoupled roll-pitch sensing, and implement a high-low level hierarchical control scheme. We then develop a particle-spring model of the handkerchief for control-oriented abstraction and strategy evaluation. Hardware experiments validate this framework, achieving an unfolding ratio of approximately 99% and fingertip tracking error of RMSE = 2.88 mm in high-dynamic spinning. These results demonstrate that integrating control-oriented modeling with a task-tailored dexterous wrist enables robust rest-to-steady-state transitions and precise periodic manipulation of highly flexible objects. More visualizations: https://slowly1113.github.io/icra2026-handkerchief/

Xiaoting Yin, Hao Shi, Yuhan Bao et al.

Achieving 3D reconstruction from images captured under optimal conditions has been extensively studied in the vision and imaging fields. However, in real-world scenarios, challenges such as motion blur and insufficient illumination often limit the performance of standard frame-based cameras in delivering high-quality images. To address these limitations, we incorporate a transmittance adjustment device at the hardware level, enabling event cameras to capture both motion and exposure events for diverse 3D reconstruction scenarios. Motion events (triggered by camera or object movement) are collected in fast-motion scenarios when the device is inactive, while exposure events (generated through controlled camera exposure) are captured during slower motion to reconstruct grayscale images for high-quality training and optimization of event-based 3D Gaussian Splatting (3DGS). Our framework supports three modes: High-Quality Reconstruction using exposure events, Fast Reconstruction relying on motion events, and Balanced Hybrid optimizing with initial exposure events followed by high-speed motion events. On the EventNeRF dataset, we demonstrate that exposure events significantly improve fine detail reconstruction compared to motion events and outperform frame-based cameras under challenging conditions such as low illumination and overexposure. Furthermore, we introduce EME-3D, a real-world 3D dataset with exposure events, motion events, camera calibration parameters, and sparse point clouds. Our method achieves faster and higher-quality reconstruction than event-based NeRF and is more cost-effective than methods combining event and RGB data. E-3DGS sets a new benchmark for event-based 3D reconstruction with robust performance in challenging conditions and lower hardware demands. The source code and dataset will be available at https://github.com/MasterHow/E-3DGS.

Zhenshan Bing, Yuan Meng, Yuqi Yun et al.

Generative model-based deep clustering frameworks excel in classifying complex data, but are limited in handling dynamic and complex features because they require prior knowledge of the number of clusters. In this paper, we propose a nonparametric deep clustering framework that employs an infinite mixture of Gaussians as a prior. Our framework utilizes a memoized online variational inference method that enables the "birth" and "merge" moves of clusters, allowing our framework to cluster data in a "dynamic-adaptive" manner, without requiring prior knowledge of the number of features. We name the framework as DIVA, a Dirichlet Process-based Incremental deep clustering framework via Variational Auto-Encoder. Our framework, which outperforms state-of-the-art baselines, exhibits superior performance in classifying complex data with dynamically changing features, particularly in the case of incremental features. We released our source code implementation at: https://github.com/Ghiara/diva

Lei Zhang, Kaixin Bai, Guowen Huang et al.

The deep learning models has significantly advanced dexterous manipulation techniques for multi-fingered hand grasping. However, the contact information-guided grasping in cluttered environments remains largely underexplored. To address this gap, we have developed a method for generating multi-fingered hand grasp samples in cluttered settings through contact semantic map. We introduce a contact semantic conditional variational autoencoder network (CoSe-CVAE) for creating comprehensive contact semantic map from object point cloud. We utilize grasp detection method to estimate hand grasp poses from the contact semantic map. Finally, an unified grasp evaluation model PointNetGPD++ is designed to assess grasp quality and collision probability, substantially improving the reliability of identifying optimal grasps in cluttered scenarios. Our grasp generation method has demonstrated remarkable success, outperforming state-of-the-art methods by at least 4.65% with 81.0% average grasping success rate in real-world single-object environment and 75.3% grasping success rate in cluttered scenes. We also proposed the multi-modal multi-fingered grasping dataset generation method. Our multi-fingered hand grasping dataset outperforms previous datasets in scene diversity, modality diversity. The dataset, code and supplementary materials can be found at https://sites.google.com/view/contact-dexnet.

Yu Zhang, Guangyao Tian, Long Wen et al.

This paper proposes a LiDAR-based goal-seeking and exploration framework, addressing the efficiency of online obstacle avoidance in unstructured environments populated with static and moving obstacles. This framework addresses two significant challenges associated with traditional dynamic control barrier functions (D-CBFs): their online construction and the diminished real-time performance caused by utilizing multiple D-CBFs. To tackle the first challenge, the framework's perception component begins with clustering point clouds via the DBSCAN algorithm, followed by encapsulating these clusters with the minimum bounding ellipses (MBEs) algorithm to create elliptical representations. By comparing the current state of MBEs with those stored from previous moments, the differentiation between static and dynamic obstacles is realized, and the Kalman filter is utilized to predict the movements of the latter. Such analysis facilitates the D-CBF's online construction for each MBE. To tackle the second challenge, we introduce buffer zones, generating Type-II D-CBFs online for each identified obstacle. Utilizing these buffer zones as activation areas substantially reduces the number of D-CBFs that need to be activated. Upon entering these buffer zones, the system prioritizes safety, autonomously navigating safe paths, and hence referred to as the exploration mode. Exiting these buffer zones triggers the system's transition to goal-seeking mode. We demonstrate that the system's states under this framework achieve safety and asymptotic stabilization. Experimental results in simulated and real-world environments have validated our framework's capability, allowing a LiDAR-equipped mobile robot to efficiently and safely reach the desired location within dynamic environments containing multiple obstacles.

Erdi Sayar, Zhenshan Bing, Carlo D'Eramo et al.

Multi-goal robot manipulation tasks with sparse rewards are difficult for reinforcement learning (RL) algorithms due to the inefficiency in collecting successful experiences. Recent algorithms such as Hindsight Experience Replay (HER) expedite learning by taking advantage of failed trajectories and replacing the desired goal with one of the achieved states so that any failed trajectory can be utilized as a contribution to learning. However, HER uniformly chooses failed trajectories, without taking into account which ones might be the most valuable for learning. In this paper, we address this problem and propose a novel approach Contact Energy Based Prioritization~(CEBP) to select the samples from the replay buffer based on rich information due to contact, leveraging the touch sensors in the gripper of the robot and object displacement. Our prioritization scheme favors sampling of contact-rich experiences, which are arguably the ones providing the largest amount of information. We evaluate our proposed approach on various sparse reward robotic tasks and compare them with the state-of-the-art methods. We show that our method surpasses or performs on par with those methods on robot manipulation tasks. Finally, we deploy the trained policy from our method to a real Franka robot for a pick-and-place task. We observe that the robot can solve the task successfully. The videos and code are publicly available at: https://erdiphd.github.io/HER_force

Hang Li, Fengyi Shen, Dong Chen et al.

Vision-language-action (VLA) models for closed-loop robot control are typically cast under the Markov assumption, making them prone to errors on tasks requiring historical context. To incorporate memory, existing VLAs either retrieve from a memory bank, which can be misled by distractors, or extend the frame window, whose fixed horizon still limits long-term retention. In this paper, we introduce ReMem-VLA, a Recurrent Memory VLA model equipped with two sets of learnable queries: frame-level recurrent memory queries for propagating information across consecutive frames to support short-term memory, and chunk-level recurrent memory queries for carrying context across temporal chunks for long-term memory. These queries are trained end-to-end to aggregate and maintain relevant context over time, implicitly guiding the model's decisions without additional training or inference cost. Furthermore, to enhance visual memory, we introduce Past Observation Prediction as an auxiliary training objective. Through extensive memory-centric simulation and real-world robot experiments, we demonstrate that ReMem-VLA exhibits strong memory capabilities across multiple dimensions, including spatial, sequential, episodic, temporal, and visual memory. ReMem-VLA significantly outperforms memory-free VLA baselines $π$0.5 and OpenVLA-OFT and surpasses MemoryVLA on memory-dependent tasks by a large margin.

Yu Zhang, Long Wen, Xiangtong Yao et al.

This paper presents an adaptive online learning framework for systems with uncertain parameters to ensure safety-critical control in non-stationary environments. Our approach consists of two phases. The initial phase is centered on a novel sparse Gaussian process (GP) framework. We first integrate a forgetting factor to refine a variational sparse GP algorithm, thus enhancing its adaptability. Subsequently, the hyperparameters of the Gaussian model are trained with a specially compound kernel, and the Gaussian model's online inferential capability and computational efficiency are strengthened by updating a solitary inducing point derived from new samples, in conjunction with the learned hyperparameters. In the second phase, we propose a safety filter based on high-order control barrier functions (HOCBFs), synergized with the previously trained learning model. By leveraging the compound kernel from the first phase, we effectively address the inherent limitations of GPs in handling high-dimensional problems for real-time applications. The derived controller ensures a rigorous lower bound on the probability of satisfying the safety specification. Finally, the efficacy of our proposed algorithm is demonstrated through real-time obstacle avoidance experiments executed using both a simulation platform and a real-world 7-DOF robot.

Hongli Xu, Lei Zhang, Xiaoyue Hu et al.

General-purpose robotic skills from end-to-end demonstrations often leads to task-specific policies that fail to generalize beyond the training distribution. Therefore, we introduce FunCanon, a framework that converts long-horizon manipulation tasks into sequences of action chunks, each defined by an actor, verb, and object. These chunks focus policy learning on the actions themselves, rather than isolated tasks, enabling compositionality and reuse. To make policies pose-aware and category-general, we perform functional object canonicalization for functional alignment and automatic manipulation trajectory transfer, mapping objects into shared functional frames using affordance cues from large vision language models. An object centric and action centric diffusion policy FuncDiffuser trained on this aligned data naturally respects object affordances and poses, simplifying learning and improving generalization ability. Experiments on simulated and real-world benchmarks demonstrate category-level generalization, cross-task behavior reuse, and robust sim2real deployment, showing that functional canonicalization provides a strong inductive bias for scalable imitation learning in complex manipulation domains. Details of the demo and supplemental material are available on our project website https://sites.google.com/view/funcanon.

Ju Dong, Lei Zhang, Liding Zhang et al.

Mobile manipulation requires the coordinated control of a mobile base and a robotic arm while simultaneously perceiving both global scene context and fine-grained object details. Existing single-view approaches often fail in unstructured environments due to limited fields of view, exploration, and generalization abilities. Moreover, classical controllers, although stable, struggle with efficiency and manipulability near singularities. To address these challenges, we propose M4Diffuser, a hybrid framework that integrates a Multi-View Diffusion Policy with a novel Reduced and Manipulability-aware QP (ReM-QP) controller for mobile manipulation. The diffusion policy leverages proprioceptive states and complementary camera perspectives with both close-range object details and global scene context to generate task-relevant end-effector goals in the world frame. These high-level goals are then executed by the ReM-QP controller, which eliminates slack variables for computational efficiency and incorporates manipulability-aware preferences for robustness near singularities. Comprehensive experiments in simulation and real-world environments show that M4Diffuser achieves 7 to 56 percent higher success rates and reduces collisions by 3 to 31 percent over baselines. Our approach demonstrates robust performance for smooth whole-body coordination, and strong generalization to unseen tasks, paving the way for reliable mobile manipulation in unstructured environments. Details of the demo and supplemental material are available on our project website https://sites.google.com/view/m4diffuser.

Yuan Meng, Xiangtong Yao, Haihui Ye et al.

Embodied long-horizon manipulation requires robotic systems to process multimodal inputs-such as vision and natural language-and translate them into executable actions. However, existing learning-based approaches often depend on large, task-specific datasets and struggle to generalize to unseen scenarios. Recent methods have explored using large language models (LLMs) as high-level planners that decompose tasks into subtasks using natural language and guide pretrained low-level controllers. Yet, these approaches assume perfect execution from low-level policies, which is unrealistic in real-world environments with noise or suboptimal behaviors. To overcome this, we fully discard the pretrained low-level policy and instead use the LLM to directly generate executable code plans within a closed-loop framework. Our planner employs chain-of-thought (CoT)-guided few-shot learning with incrementally structured examples to produce robust and generalizable task plans. Complementing this, a reporter evaluates outcomes using RGB-D and delivers structured feedback, enabling recovery from misalignment and replanning under partial observability. This design eliminates per-step inference, reduces computational overhead, and limits error accumulation that was observed in previous methods. Our framework achieves state-of-the-art performance on 30+ diverse seen and unseen long-horizon tasks across LoHoRavens, CALVIN, Franka Kitchen, and cluttered real-world settings.

Yuan Meng, Xiangtong Yao, Kejia Chen et al.

Reinforcement learning (RL) methods typically learn new tasks from scratch, often disregarding prior knowledge that could accelerate the learning process. While some methods incorporate previously learned skills, they usually rely on a fixed structure, such as a single Gaussian distribution, to define skill priors. This rigid assumption can restrict the diversity and flexibility of skills, particularly in complex, long-horizon tasks. In this work, we introduce a method that models potential primitive skill motions as having non-parametric properties with an unknown number of underlying features. We utilize a Bayesian non-parametric model, specifically Dirichlet Process Mixtures, enhanced with birth and merge heuristics, to pre-train a skill prior that effectively captures the diverse nature of skills. Additionally, the learned skills are explicitly trackable within the prior space, enhancing interpretability and control. By integrating this flexible skill prior into an RL framework, our approach surpasses existing methods in long-horizon manipulation tasks, enabling more efficient skill transfer and task success in complex environments. Our findings show that a richer, non-parametric representation of skill priors significantly improves both the learning and execution of challenging robotic tasks. All data, code, and videos are available at https://ghiara.github.io/HELIOS/.

Haitao Meng, Chonghao Zhong, Sheng Tang et al.

Event cameras are neuromorphically inspired sensors that sparsely and asynchronously report brightness changes. Their unique characteristics of high temporal resolution, high dynamic range, and low power consumption make them well-suited for addressing challenges in monocular depth estimation (e.g., high-speed or low-lighting conditions). However, current existing methods primarily treat event streams as black-box learning systems without incorporating prior physical principles, thus becoming over-parameterized and failing to fully exploit the rich temporal information inherent in event camera data. To address this limitation, we incorporate physical motion principles to propose an interpretable monocular depth estimation framework, where the likelihood of various depth hypotheses is explicitly determined by the effect of motion compensation. To achieve this, we propose a Focus Cost Discrimination (FCD) module that measures the clarity of edges as an essential indicator of focus level and integrates spatial surroundings to facilitate cost estimation. Furthermore, we analyze the noise patterns within our framework and improve it with the newly introduced Inter-Hypotheses Cost Aggregation (IHCA) module, where the cost volume is refined through cost trend prediction and multi-scale cost consistency constraints. Extensive experiments on real-world and synthetic datasets demonstrate that our proposed framework outperforms cutting-edge methods by up to 10\% in terms of the absolute relative error metric, revealing superior performance in predicting accuracy.

Yu Zhang, Haoan Ping, Yuchen Li et al.

Recent salient object detection (SOD) methods aim to improve performance in four key directions: semantic enhancement, boundary refinement, auxiliary task supervision, and multi-modal fusion. In pursuit of continuous gains, these approaches have evolved toward increasingly sophisticated architectures with multi-stage pipelines, specialized fusion modules, edge-guided learning, and elaborate attention mechanisms. However, this complexity paradoxically introduces feature redundancy and cross-component interference that obscure salient cues, ultimately reaching performance bottlenecks. In contrast, human vision achieves efficient salient object identification without such architectural complexity. This contrast raises a fundamental question: can we design a biologically grounded yet architecturally simple SOD framework that dispenses with most of this engineering complexity, while achieving state-of-the-art accuracy, computational efficiency, and interpretability? In this work, we answer this question affirmatively by introducing DualGazeNet, a biologically inspired pure Transformer framework that models the dual biological principles of robust representation learning and magnocellular-parvocellular dual-pathway processing with cortical attention modulation in the human visual system. Extensive experiments on five RGB SOD benchmarks show that DualGazeNet consistently surpasses 25 state-of-the-art CNN- and Transformer-based methods. On average, DualGazeNet achieves about 60\% higher inference speed and 53.4\% fewer FLOPs than four Transformer-based baselines of similar capacity (VST++, MDSAM, Sam2unet, and BiRefNet). Moreover, DualGazeNet exhibits strong cross-domain generalization, achieving leading or highly competitive performance on camouflaged and underwater SOD benchmarks without relying on additional modalities.

Zhenguo Sun, Yibo Peng, Yuan Meng et al.

Long-horizon, high-dynamic motion tracking on humanoids remains brittle because absolute joint commands cannot compensate model-plant mismatch, leading to error accumulation. We propose RobotDancing, a simple, scalable framework that predicts residual joint targets to explicitly correct dynamics discrepancies. The pipeline is end-to-end--training, sim-to-sim validation, and zero-shot sim-to-real--and uses a single-stage reinforcement learning (RL) setup with a unified observation, reward, and hyperparameter configuration. We evaluate primarily on Unitree G1 with retargeted LAFAN1 dance sequences and validate transfer on H1/H1-2. RobotDancing can track multi-minute, high-energy behaviors (jumps, spins, cartwheels) and deploys zero-shot to hardware with high motion tracking quality.

Hongkuan Zhou, Zhenshan Bing, Xiangtong Yao et al.

The growing interest in language-conditioned robot manipulation aims to develop robots capable of understanding and executing complex tasks, with the objective of enabling robots to interpret language commands and manipulate objects accordingly. While language-conditioned approaches demonstrate impressive capabilities for addressing tasks in familiar environments, they encounter limitations in adapting to unfamiliar environment settings. In this study, we propose a general-purpose, language-conditioned approach that combines base skill priors and imitation learning under unstructured data to enhance the algorithm's generalization in adapting to unfamiliar environments. We assess our model's performance in both simulated and real-world environments using a zero-shot setting. In the simulated environment, the proposed approach surpasses previously reported scores for CALVIN benchmark, especially in the challenging Zero-Shot Multi-Environment setting. The average completed task length, indicating the average number of tasks the agent can continuously complete, improves more than 2.5 times compared to the state-of-the-art method HULC. In addition, we conduct a zero-shot evaluation of our policy in a real-world setting, following training exclusively in simulated environments without additional specific adaptations. In this evaluation, we set up ten tasks and achieved an average 30% improvement in our approach compared to the current state-of-the-art approach, demonstrating a high generalization capability in both simulated environments and the real world. For further details, including access to our code and videos, please refer to https://hk-zh.github.io/spil/

Christian Creß, Zhenshan Bing, Alois C. Knoll

The main problems in transportation are accidents, increasingly slow traffic flow, and pollution. An intelligent transportation system (ITS) using external infrastructure can overcome these problems. For this reason, the number of such systems is increasing dramatically, and therefore requires an adequate overview. To the best of our knowledge, no current systematic review of existing ITS solutions exists. To fill this knowledge gap, our paper provides an overview of existing ITS that use external infrastructure worldwide. Accordingly, this paper addresses current questions and challenges. For this purpose, we performed a literature review of documents that describe existing ITS solutions from 2009 until today. We categorized the results according to technology levels and analyzed its hardware system setup and value-added contributions. In doing so, we made the ITS solutions comparable and highlighted past development alongside current trends. We analyzed more than 357 papers, including 52 test bed projects. In summary, current ITSs can deliver accurate information about individuals in traffic situations in real-time. However, further research into ITS should focus on more reliable perception of the traffic using modern sensors, plug-and-play mechanisms, and secure real-time distribution of the digital twins in a decentralized manner. By addressing these topics, the development of intelligent transportation systems will be able to take a step towards its comprehensive roll-out.

Genghang Zhuang, Carlo Cagnetta, Zhenshan Bing et al.

Localization in the environment is an essential navigational capability for animals and mobile robots. In the indoor environment, the global localization problem remains challenging to be perfectly solved with probabilistic methods. However, animals are able to instinctively localize themselves with much less effort. Therefore, it is intriguing and promising to seek biological inspiration from animals. In this paper, we present a biologically-inspired global localization system using a LiDAR sensor that utilizes a hippocampal model and a landmark-based re-localization approach. The experiment results show that the proposed method is competitive to Monte Carlo Localization, and the results demonstrate the high accuracy, applicability, and reliability of the proposed biologically-inspired localization system in different localization scenarios.

Genghang Zhuang, Zhenshan Bing, Yuhong Huang et al.

Simultaneous localization and mapping (SLAM) is one of the essential techniques and functionalities used by robots to perform autonomous navigation tasks. Inspired by the rodent hippocampus, this paper presents a biologically inspired SLAM system based on a LiDAR sensor using a hippocampal model to build a cognitive map and estimate the robot pose in indoor environments. Based on the biologically inspired models mimicking boundary cells, place cells, and head direction cells, the SLAM system using LiDAR point cloud data is capable of leveraging the self-motion cues from the LiDAR odometry and the boundary cues from the LiDAR boundary cells to build a cognitive map and estimate the robot pose. Experiment results show that with the LiDAR boundary cells the proposed SLAM system greatly outperforms the camera-based brain-inspired method in both simulation and indoor environments, and is competitive with the conventional LiDAR-based SLAM methods.

Zhenshan Bing, Amir EI Sewisy, Genghang Zhuang et al.

As a vital cognitive function of animals, the navigation skill is first built on the accurate perception of the directional heading in the environment. Head direction cells (HDCs), found in the limbic system of animals, are proven to play an important role in identifying the directional heading allocentrically in the horizontal plane, independent of the animal's location and the ambient conditions of the environment. However, practical HDC models that can be implemented in robotic applications are rarely investigated, especially those that are biologically plausible and yet applicable to the real world. In this paper, we propose a computational HDC network which is consistent with several neurophysiological findings concerning biological HDCs, and then implement it in robotic navigation tasks. The HDC network keeps a representation of the directional heading only relying on the angular velocity as an input. We examine the proposed HDC model in extensive simulations and real-world experiments and demonstrate its excellent performance in terms of accuracy and real-time capability.

Zhenshan Bing, Lukas Knak, Fabrice Oliver Robin et al.

Recent state-of-the-art artificial agents lack the ability to adapt rapidly to new tasks, as they are trained exclusively for specific objectives and require massive amounts of interaction to learn new skills. Meta-reinforcement learning (meta-RL) addresses this challenge by leveraging knowledge learned from training tasks to perform well in previously unseen tasks. However, current meta-RL approaches limit themselves to narrow parametric task distributions, ignoring qualitative differences between tasks that occur in the real world. In this paper, we introduce TIGR, a Task-Inference-based meta-RL algorithm using Gaussian mixture models (GMM) and gated Recurrent units, designed for tasks in non-parametric environments. We employ a generative model involving a GMM to capture the multi-modality of the tasks. We decouple the policy training from the task-inference learning and efficiently train the inference mechanism on the basis of an unsupervised reconstruction objective. We provide a benchmark with qualitatively distinct tasks based on the half-cheetah environment and demonstrate the superior performance of TIGR compared to state-of-the-art meta-RL approaches in terms of sample efficiency (3-10 times faster), asymptotic performance, and applicability in non-parametric environments with zero-shot adaptation.

Zhenshan Bing, Matthias Brucker, Fabrice O. Morin et al.

Reinforcement learning algorithms such as hindsight experience replay (HER) and hindsight goal generation (HGG) have been able to solve challenging robotic manipulation tasks in multi-goal settings with sparse rewards. HER achieves its training success through hindsight replays of past experience with heuristic goals, but under-performs in challenging tasks in which goals are difficult to explore. HGG enhances HER by selecting intermediate goals that are easy to achieve in the short term and promising to lead to target goals in the long term. This guided exploration makes HGG applicable to tasks in which target goals are far away from the object's initial position. However, HGG is not applicable to manipulation tasks with obstacles because the euclidean metric used for HGG is not an accurate distance metric in such environments. In this paper, we propose graph-based hindsight goal generation (G-HGG), an extension of HGG selecting hindsight goals based on shortest distances in an obstacle-avoiding graph, which is a discrete representation of the environment. We evaluated G-HGG on four challenging manipulation tasks with obstacles, where significant enhancements in both sample efficiency and overall success rate are shown over HGG and HER. Videos can be viewed at https://sites.google.com/view/demos-g-hgg/.

Zhenshan Bing, Claus Meschede, Guang Chen et al.

Building spiking neural networks (SNNs) based on biological synaptic plasticities holds a promising potential for accomplishing fast and energy-efficient computing, which is beneficial to mobile robotic applications. However, the implementations of SNNs in robotic fields are limited due to the lack of practical training methods. In this paper, we therefore introduce both indirect and direct end-to-end training methods of SNNs for a lane-keeping vehicle. First, we adopt a policy learned using the \textcolor{black}{Deep Q-Learning} (DQN) algorithm and then subsequently transfer it to an SNN using supervised learning. Second, we adopt the reward-modulated spike-timing-dependent plasticity (R-STDP) for training SNNs directly, since it combines the advantages of both reinforcement learning and the well-known spike-timing-dependent plasticity (STDP). We examine the proposed approaches in three scenarios in which a robot is controlled to keep within lane markings by using an event-based neuromorphic vision sensor. We further demonstrate the advantages of the R-STDP approach in terms of the lateral localization accuracy and training time steps by comparing them with other three algorithms presented in this paper.

Zhenshan Bing, Christian Lemke, Zhuangyi Jiang et al.

Similar to their counterparts in nature, the flexible bodies of snake-like robots enhance their movement capability and adaptability in diverse environments. However, this flexibility corresponds to a complex control task involving highly redundant degrees of freedom, where traditional model-based methods usually fail to propel the robots energy-efficiently. In this work, we present a novel approach for designing an energy-efficient slithering gait for a snake-like robot using a model-free reinforcement learning (RL) algorithm. Specifically, we present an RL-based controller for generating locomotion gaits at a wide range of velocities, which is trained using the proximal policy optimization (PPO) algorithm. Meanwhile, a traditional parameterized gait controller is presented and the parameter sets are optimized using the grid search and Bayesian optimization algorithms for the purposes of reasonable comparisons. Based on the analysis of the simulation results, we demonstrate that this RL-based controller exhibits very natural and adaptive movements, which are also substantially more energy-efficient than the gaits generated by the parameterized controller. Videos are shown at \textcolor{blue}{\href{https://videoviewsite.wixsite.com/rlsnake}{https://videoviewsite.wixsite.com/rlsnake}}.