CoRL 2024 论文列表

Human hands are capable of in-hand manipulation in the presence of different hand motions. For a robot hand, harnessing rich tactile information to achieve this level of dexterity still remains a significant challenge. In this paper, we present AnyRotate, a system for gravity-invariant multi-axis in-hand object rotation using dense featured sim-to-real touch. We tackle this problem by training a dense tactile policy in simulation and present a sim-to-real method for rich tactile sensing to achieve zero-shot policy transfer. Our formulation allows the training of a unified policy to rotate unseen objects about arbitrary rotation axes in any hand direction. In our experiments, we highlight the benefit of capturing detailed contact information when handling objects of varying properties. Interestingly, we found rich multi-fingered tactile sensing can detect unstable grasps and provide a reactive behavior that improves the robustness of the policy.

We present Splat-MOVER, a modular robotics stack for open-vocabulary robotic manipulation, which leverages the editability of Gaussian Splatting (GSplat) scene representations to enable multi-stage manipulation tasks. Splat-MOVER consists of: (i) ASK-Splat, a GSplat representation that distills semantic and grasp affordance features into the 3D scene. ASK-Splat enables geometric, semantic, and affordance understanding of 3D scenes, which is critical for many robotics tasks; (ii) SEE-Splat, a real-time scene-editing module using 3D semantic masking and infilling to visualize the motions of objects that result from robot interactions in the real-world. SEE-Splat creates a “digital twin” of the evolving environment throughout the manipulation task; and (iii) Grasp-Splat, a grasp generation module that uses ASK-Splat and SEE-Splat to propose affordance-aligned candidate grasps for open-world objects. ASK-Splat is trained in real-time from RGB images in a brief scanning phase prior to operation, while SEE-Splat and Grasp-Splat run in real-time during operation. We demonstrate the superior performance of Splat-MOVER in hardware experiments on a Kinova robot compared to two recent baselines in four single-stage, open-vocabulary manipulation tasks. In addition, we demonstrate Splat-MOVER in four multi-stage manipulation tasks, using the edited scene to reflect changes due to prior manipulation stages, which is not possible with existing baselines. Video demonstrations and the code for the project are available at https://splatmover.github.io.

Learning to move is a primary goal for animals and robots, where ensuring safety is often important when optimizing control policies on the embodied systems. For complex tasks such as the control of human or humanoid control, the high-dimensional parameter space adds complexity to the safe optimization effort. Current safe exploration algorithms exhibit inefficiency and may even become infeasible with large high-dimensional input spaces. Furthermore, existing high-dimensional constrained optimization methods neglect safety in the search process. In this paper, we propose High-dimensional Safe Bayesian Optimization with local optimistic exploration (HdSafeBO), a novel approach designed to handle high-dimensional sampling problems under probabilistic safety constraints. We introduce a local optimistic strategy to efficiently and safely optimize the objective function, providing a probabilistic safety guarantee and a cumulative safety violation bound. Through the use of isometric embedding, HdSafeBO addresses problems ranging from a few hundred to several thousand dimensions while maintaining safety guarantees. To our knowledge, HdSafeBO is the first algorithm capable of optimizing the control of high-dimensional musculoskeletal systems with high safety probability. We also demonstrate the real-world applicability of HdSafeBO through its use in the safe online optimization of neural stimulation-induced human motion control.

This work proposes a retrieve-and-transfer framework for zero-shot robotic manipulation, dubbed RAM, featuring generalizability across various objects, environments, and embodiments. Unlike existing approaches that learn manipulation from expensive in-domain demonstrations, RAM capitalizes on a retrieval-based affordance transfer paradigm to acquire versatile manipulation capabilities from abundant out-of-domain data. RAM first extracts unified affordance at scale from diverse sources of demonstrations including robotic data, human-object interaction (HOI) data, and custom data to construct a comprehensive affordance memory. Then given a language instruction, RAM hierarchically retrieves the most similar demonstration from the affordance memory and transfers such out-of-domain 2D affordance to in-domain 3D actionable affordance in a zero-shot and embodiment-agnostic manner. Extensive simulation and real-world evaluations demonstrate that our RAM consistently outperforms existing works in diverse daily tasks. Additionally, RAM shows significant potential for downstream applications such as automatic and efficient data collection, one-shot visual imitation, and LLM/VLM-integrated long-horizon manipulation.

Control Barrier Functions (CBFs) offer an elegant framework for constraining nonlinear control system dynamics to an invariant subset of a pre-specified safe set. However, finding a CBF that simultaneously promotes performance by maximizing the resulting control invariant set while accommodating complex safety constraints, especially in high relative degree systems with actuation constraints, remains a significant challenge. In this work, we propose a novel self-supervised learning framework that holistically addresses these hurdles. Given a Boolean composition of multiple state constraints defining the safe set, our approach begins by constructing a smooth function whose zero superlevel set provides an inner approximation of the safe set. This function is then used with a smooth neural network to parameterize the CBF candidate. Finally, we design a physics-informed training loss function based on a Hamilton-Jacobi Partial Differential Equation (PDE) to train the PINN-CBF and enlarge the volume of the induced control invariant set. We demonstrate the effectiveness of our approach on a 2D double integrator (DI) system and a 7D fixed-wing aircraft system (F16).

Intelligent robots capable of improving from autonomously collected experience have the potential to transform robot learning: instead of collecting costly teleoperated demonstration data, large-scale deployment of fleets of robots can quickly collect larger quantities of autonomous data useful for training better robot policies. However, autonomous improvement requires solving two key problems: (i) fully automating a scalable data collection procedure that can collect diverse and semantically meaningful robot data and (ii) learning from non-optimal, autonomous data with no human annotations. To this end, we propose a novel approach that addresses these challenges, allowing instruction following policies to improve from autonomously collected data without human supervision. Our framework leverages vision-language models to collect and evaluate semantically meaningful experiences in new environments, and then utilizes a decomposition of instruction following tasks into (semantic) language-conditioned image generation and (non-semantic) goal reaching, which makes it significantly more practical to improve from this autonomously collected data without any human annotations. We carry out extensive experiments in the real world to demonstrate the effectiveness of our approach, and find that in a suite of unseen environments, the robot policy can be improved significantly with autonomously collected data. We open-source the code for our semantic autonomous improvement pipeline, as well as our autonomous dataset of 25K trajectories collected across five tabletop environments: https://soar-autonomous-improvement.github.io

We present the first static-obstacle avoidance method for quadrotors using just an onboard, monocular event camera. Quadrotors are capable of fast and agile flight in cluttered environments when piloted manually, but vision-based autonomous flight in unknown environments is difficult in part due to the sensor limitations of traditional onboard cameras. Event cameras, however, promise nearly zero motion blur and high dynamic range, but produce a very large volume of events under significant ego-motion and further lack a continuous-time sensor model in simulation, making direct sim-to-real transfer not possible. By leveraging depth prediction as a pretext task in our learning framework, we can pre-train a reactive obstacle avoidance events-to-control policy with approximated, simulated events and then fine-tune the perception component with limited events-and-depth real-world data to achieve obstacle avoidance in indoor and outdoor settings. We demonstrate this across two quadrotor-event camera platforms in multiple settings and find, contrary to traditional vision-based works, that low speeds (1m/s) make the task harder and more prone to collisions, while high speeds (5m/s) result in better event-based depth estimation and avoidance. We also find that success rates in outdoor scenes can be significantly higher than in certain indoor scenes.

Tactile sensing is critical to fine-grained, contact-rich manipulation tasks, such as insertion and assembly. Prior research has shown the possibility of learning tactile-guided policy from teleoperated demonstration data. However, to provide the demonstration, human users often rely on visual feedback to control the robot. This creates a gap between the sensing modality used for controlling the robot (visual) and the modality of interest (tactile). To bridge this gap, we introduce "MimicTouch'', a novel framework for learning policies directly from demonstrations provided by human users with their hands. The key innovations are i) a human tactile data collection system which collects multi-modal tactile dataset for learning human's tactile-guided control strategy, ii) an imitation learning-based framework for learning human's tactile-guided control strategy through such data, and iii) an online residual RL framework to bridge the embodiment gap between the human hand and the robot gripper. Through comprehensive experiments, we highlight the efficacy of utilizing human's tactile-guided control strategy to resolve contact-rich manipulation tasks. The project website is at https://sites.google.com/view/MimicTouch.

Diffusion-based policies have shown remarkable capability in executing complex robotic manipulation tasks but lack explicit characterization of geometry and semantics, which often limits their ability to generalize to unseen objects and layouts. To enhance the generalization capabilities of Diffusion Policy, we introduce a novel framework that incorporates explicit spatial and semantic information via 3D semantic fields. We generate 3D descriptor fields from multi-view RGBD observations with large foundational vision models, then compare these descriptor fields against reference descriptors to obtain semantic fields. The proposed method explicitly considers geometry and semantics, enabling strong generalization capabilities in tasks requiring category-level generalization, resolving geometric ambiguities, and attention to subtle geometric details. We evaluate our method across eight tasks involving articulated objects and instances with varying shapes and textures from multiple object categories. Our method demonstrates its effectiveness by increasing Diffusion Policy's average success rate on \textit{unseen} instances from 20\% to 93\%. Additionally, we provide a detailed analysis and visualization to interpret the sources of performance gain and explain how our method can generalize to novel instances. Project page: https://robopil.github.io/GenDP/

It is desired to equip robots with the capability of interacting with various soft materials as they are ubiquitous in the real world. While physics simulations are one of the predominant methods for data collection and robot training, simulating soft materials presents considerable challenges. Specifically, it is significantly more costly than simulating rigid objects in terms of simulation speed and storage requirements. These limitations typically restrict the scope of studies on soft materials to small and bounded areas, thereby hindering the learning of skills in broader spaces. To address this issue, we introduce UBSoft, a new simulation platform designed to support unbounded soft environments for robot skill acquisition. Our platform utilizes spatially adaptive resolution scales, where simulation resolution dynamically adjusts based on proximity to active robotic agents. Our framework markedly reduces the demand for extensive storage space and computation costs required for large-scale scenarios involving soft materials. We also establish a set of benchmark tasks in our platform, including both locomotion and manipulation tasks, and conduct experiments to evaluate the efficacy of various reinforcement learning algorithms and trajectory optimization techniques, both gradient-based and sampling-based. Preliminary results indicate that sampling-based trajectory optimization generally achieves better results for obtaining one trajectory to solve the task. Additionally, we conduct experiments in real-world environments to demonstrate that advancements made in our UBSoft simulator could translate to improved robot interactions with large-scale soft material. More videos can be found at https://ubsoft24.github.io.

Bimanual robotic manipulation with dexterous hands has a large potential workability and a wide workspace as it follows the most natural human workflow. Learning from human demonstrations has proven highly effective for learning a dexterous manipulation policy. To collect such data, teleoperation serves as a straightforward and efficient way to do so. However, a cost-effective and easy-to-use teleoperation system is lacking for anthropomorphic robot hands. To fill the deficiency, we developed \our, a cross-platform visual-exoskeleton system for low-cost dexterous teleoperation. Our system employs a hand-facing camera to capture 3D hand poses and an exoskeleton mounted on a base that can be easily carried on users' backs. ACE captures both the hand root end-effector and hand pose in real-time and enables cross-platform operations. We evaluate the key system parameters compared with previous teleoperation systems and show clear advantages of \our. We then showcase the desktop and mobile versions of our system on six different robot platforms (including humanoid-hands, arm-hands, arm-gripper, and quadruped-gripper systems), and demonstrate the effectiveness of learning three difficult real-world tasks through the collected demonstration on two of them.

Training robot policies in the real world can be unsafe, costly, and difficult to scale. Simulation serves as an inexpensive and potentially limitless source of training data, but suffers from the semantics and physics disparity between simulated and real-world environments. These discrepancies can be minimized by training in *digital twins*, which serve as virtual replicas of a real scene but are expensive to generate and cannot produce cross-domain generalization. To address these limitations, we propose the concept of ***digital cousins***, a virtual asset or scene that, unlike a *digital twin*, does not explicitly model a real-world counterpart but still exhibits similar geometric and semantic affordances. As a result, *digital cousins* simultaneously reduce the cost of generating an analogous virtual environment while also facilitating better robustness during sim-to-real domain transfer by providing a distribution of similar training scenes. Leveraging digital cousins, we introduce a novel method for their automated creation, and propose a fully automated real-to-sim-to-real pipeline for generating fully interactive scenes and training robot policies that can be deployed zero-shot in the original scene. We find that digital cousin scenes that preserve geometric and semantic affordances can be produced automatically, and can be used to train policies that outperform policies trained on digital twins, achieving 90\% vs. 25\% success rates under zero-shot sim-to-real transfer. Additional details are available at https://digital-cousins.github.io/.

Depth sensing is an important problem for 3D vision-based robotics. Yet, a real-world active stereo or ToF depth camera often produces noisy and incomplete depth which bottlenecks robot performances. In this work, we propose D3RoMa, a learning-based depth estimation framework on stereo image pairs that predicts clean and accurate depth in diverse indoor scenes, even in the most challenging scenarios with translucent or specular surfaces where classical depth sensing completely fails. Key to our method is that we unify depth estimation and restoration into an image-to-image translation problem by predicting the disparity map with a denoising diffusion probabilistic model. At inference time, we further incorporated a left-right consistency constraint as classifier guidance to the diffusion process. Our framework combines recently advanced learning-based approaches and geometric constraints from traditional stereo vision. For model training, we create a large scene-level synthetic dataset with diverse transparent and specular objects to compensate for existing tabletop datasets. The trained model can be directly applied to real-world in-the-wild scenes and achieve state-of-the-art performance in multiple public depth estimation benchmarks. Further experiments in both simulated and real environments show that accurate depth prediction significantly improves robotic manipulation in various scenarios.

LiDAR-based 3D object detection is a critical technology for the development of autonomous driving and robotics. However, the high cost of data annotation limits its advancement. We propose a novel and effective active learning (AL) method called Distribution Discrepancy and Feature Heterogeneity (DDFH), which simultaneously considers geometric features and model embeddings, assessing information from both the instance-level and frame-level perspectives. Distribution Discrepancy evaluates the difference and novelty of instances within the unlabeled and labeled distributions, enabling the model to learn efficiently with limited data. Feature Heterogeneity ensures the heterogeneity of intra-frame instance features, maintaining feature diversity while avoiding redundant or similar instances, thus minimizing annotation costs. Finally, multiple indicators are efficiently aggregated using Quantile Transform, providing a unified measure of informativeness. Extensive experiments demonstrate that DDFH outperforms the current state-of-the-art (SOTA) methods on the KITTI and Waymo datasets, effectively reducing the bounding box annotation cost by 56.3% and showing robustness when working with both one-stage and two-stage models.

This paper presents a teleoperation framework designed for online learning and adaptation of tactile skills, which provides an intuitive interface without need for physical access to execution robot. The proposed tele-teaching approach utilizes periodical Dynamical Movement Primitives (DMP) and Recursive Least Square (RLS) for generating tactile skills. An autonomy allocation strategy, guided by the learning confidence and operator intention, ensures a smooth transition between human demonstration to autonomous robot operation. Our experimental results with two 7 Degree of Freedom (DoF) Franka Panda robot demonstrates that the tele-teaching framework facilitates online motion and force learning and adaptation within a few iterations.

Large, general-purpose robotic policies trained on diverse demonstration datasets have been shown to be remarkably effective both for controlling a variety of robots in a range of different scenes, and for acquiring broad repertoires of manipulation skills. However, the data that such policies are trained on is generally of mixed quality -- not only are human-collected demonstrations unlikely to perform the task perfectly, but the larger the dataset is, the harder it is to curate only the highest quality examples. It also remains unclear how optimal data from one embodiment is for training on another embodiment. In this paper, we present a general and broadly applicable approach that enhances the performance of such generalist robot policies at deployment time by re-ranking their actions according to a value function learned via offline RL. This approach, which we call Value-Guided Policy Steering (V-GPS), is compatible with a wide range of different generalist policies, without needing to fine-tune or even access the weights of the policy. We show that the same value function can improve the performance of five different state-of-the-art policies with different architectures, even though they were trained on distinct datasets, attaining consistent performance improvement on multiple robotic platforms across a total of 12 tasks. Code and videos can be found at: https://nakamotoo.github.io/V-GPS

We present the development and real world demonstration of an in-flight attitude control law for a small low-cost quadruped with a five-bar-linkage leg design using only its legs as reaction masses. The control law is trained using deep reinforcement learning (DRL) and specifically through Proximal Policy Optimization (PPO) in the NVIDIA Omniverse Isaac Sim simulator with a GPU-accelerated DRL pipeline. To demonstrate the policy, a small quadruped is designed, constructed, and evaluated both on a rotating pole test setup and in free fall. During a free fall of 0.7 seconds, the quadruped follows commanded attitude steps of 45 degrees in all principal axes, and achieves an average base angular velocity of 110 degrees per second during large attitude reference steps.

Robotic simulation today remains challenging to scale up due to the human efforts required to create diverse simulation tasks and scenes. Simulation-trained policies also face scalability issues as many sim-to-real methods focus on a single task. To address these challenges, this work proposes GenSim2, a scalable framework that leverages coding LLMs with multi-modal and reasoning capabilities for complex and realistic simulation task creation, including long-horizon tasks with articulated objects. To automatically generate demonstration data for these tasks at scale, we propose planning and RL solvers that generalize within object categories. The pipeline can generate data for up to 100 articulated tasks with 200 objects and reduce the required human efforts. To utilize such data, we propose an effective multi-task language-conditioned policy architecture, dubbed proprioceptive point-cloud transformer (PPT), that learns from the generated demonstrations and exhibits strong sim-to-real zero-shot transfer. Combining the proposed pipeline and the policy architecture, we show a promising usage of GenSim2 that the generated data can be used for zero-shot transfer or co-train with real-world collected data, which enhances the policy performance by 20% compared with training exclusively on limited real data.

Traditional methods for synthesizing 6-DoF grasp poses from 3D observations often rely on geometric heuristics, resulting in poor generalizability, limited grasp options, and higher failure rates. Recently, data-driven methods have been proposed that use generative models to learn the distribution of grasp poses and generate diverse candidate poses. The main drawback of these methods is that they fail to achieve SE(3)-equivariance, meaning that the generated grasp poses do not transform correctly with object rotations and translations. In this paper, we propose \textit{EquiGraspFlow}, a flow-based SE(3)-equivariant 6-DoF grasp pose generative model that can learn complex conditional distributions on the SE(3) manifold while guaranteeing SE(3)-equivariance. Our model achieves the equivariance without relying on data augmentation, by using network architectures that guarantee the equivariance by construction. Extensive experiments show that \textit{EquiGraspFlow} accurately learns grasp pose distribution, achieves the SE(3)-equivariance, and significantly outperforms existing grasp pose generative models. Code is available at https://github.com/bdlim99/EquiGraspFlow.