CoRL 2024 论文列表

We present RiEMann, an end-to-end near Real-time SE(3)-Equivariant Robot Manipulation imitation learning framework from scene point cloud input. Compared to previous methods that rely on descriptor field matching, RiEMann directly predicts the target actions for manipulation without any object segmentation. RiEMann can efficiently train the visuomotor policy from scratch with 5 to 10 demonstrations for a manipulation task, generalizes to unseen SE(3) transformations and instances of target objects, resists visual interference of distracting objects, and follows the near real-time pose change of the target object. The scalable SE(3)-equivariant action space of RiEMann supports both pick-and-place tasks and articulated object manipulation tasks. In simulation and real-world 6-DOF robot manipulation experiments, we test RiEMann on 5 categories of manipulation tasks with a total of 25 variants and show that RiEMann outperforms baselines in both task success rates and SE(3) geodesic distance errors (reduced by 68.6%), and achieves 5.4 frames per second (fps) network inference speed.

Reinforcement learning (RL) is a promising approach for solving robotic manipulation tasks. However, it is challenging to apply the RL algorithms directly in the real world. For one thing, RL is data-intensive and typically requires millions of interactions with environments, which are impractical in real scenarios. For another, it is necessary to make heavy engineering efforts to design reward functions manually. To address these issues, we leverage foundation models in this paper. We propose Reinforcement Learning with Foundation Priors (RLFP) to utilize guidance and feedback from policy, value, and success-reward foundation models. Within this framework, we introduce the Foundation-guided Actor-Critic (FAC) algorithm, which enables embodied agents to explore more efficiently with automatic reward functions. The benefits of our framework are threefold: (1) \textit{sample efficient}; (2) \textit{minimal and effective reward engineering}; (3) \textit{agnostic to foundation model forms and robust to noisy priors}. Our method achieves remarkable performances in various manipulation tasks on both real robots and in simulation. Across 5 dexterous tasks with real robots, FAC achieves an average success rate of 86\% after one hour of real-time learning. Across 8 tasks in the simulated Meta-world, FAC achieves 100\% success rates in 7/8 tasks under less than 100k frames (about 1-hour training), outperforming baseline methods with manual-designed rewards in 1M frames. We believe the RLFP framework can enable future robots to explore and learn autonomously in the physical world for more tasks.

Unsupervised goal-conditioned reinforcement learning (GCRL) is a promising paradigm for developing diverse robotic skills without external supervision. However, existing unsupervised GCRL methods often struggle to cover a wide range of states in complex environments due to their limited exploration and sparse or noisy rewards for GCRL. To overcome these challenges, we propose a novel unsupervised GCRL method that leverages TemporaL Distance-aware Representations (TLDR). Based on temporal distance, TLDR selects faraway goals to initiate exploration and computes intrinsic exploration rewards and goal-reaching rewards. Specifically, our exploration policy seeks states with large temporal distances (i.e. covering a large state space), while the goal-conditioned policy learns to minimize the temporal distance to the goal (i.e. reaching the goal). Our results in six simulated locomotion environments demonstrate that TLDR significantly outperforms prior unsupervised GCRL methods in achieving a wide range of states.

Robot evaluations in language-guided, real world settings are time-consuming and often sample only a small space of potential instructions across complex scenes. In this work, we introduce contrast sets for robotics as an approach to make small, but specific, perturbations to otherwise independent, identically distributed (i.i.d.) test instances. We investigate the relationship between experimenter effort to carry out an evaluation and the resulting estimated test performance as well as the insights that can be drawn from performance on perturbed instances. We use contrast sets to characterize policies at reduced experimenter effort in both a simulated manipulation task and a physical robot vision-and-language navigation task. We encourage the use of contrast set evaluations as a more informative alternative to small scale, i.i.d. demonstrations on physical robots, and as a scalable alternative to industry-scale real world evaluations.

We present MOSAIC, a modular architecture for coordinating multiple robots to (a) interact with users using natural language and (b) manipulate an open vocabulary of everyday objects. At several levels, MOSAIC employs modularity: it leverages multiple large-scale pre-trained models for high-level tasks like language and image recognition, while using streamlined modules designed for low-level task-specific control. This decomposition allows us to reap the complementary benefits of foundation models and precise, more specialized models, enabling our system to scale to complex tasks that involve coordinating multiple robots and humans. First, we unit-test individual modules with 180 episodes of visuomotor picking, 60 episodes of human motion forecasting, and 46 online user evaluations of the task planner. We then extensively evaluate MOSAIC with 60 end-to-end trials. We discuss crucial design decisions, limitations of the current system, and open challenges in this domain

The requirements for real-world manipulation tasks are diverse and often conflicting; some tasks require precise motion while others require force compliance; some tasks require avoidance of certain regions while others require convergence to certain states. Satisfying these varied requirements with a fixed state-action representation and control strategy is challenging, impeding the development of a universal robotic foundation model. In this work, we propose Meta-Control, the first LLM-enabled automatic control synthesis approach that creates customized state representations and control strategies tailored to specific tasks. Our core insight is that a meta-control system can be built to automate the thought process that human experts use to design control systems. Specifically, human experts heavily use a model-based, hierarchical (from abstract to concrete) thought model, then compose various dynamic models and controllers together to form a control system. Meta-Control mimics the thought model and harnesses LLM's extensive control knowledge with Socrates' "art of midwifery" to automate the thought process. Meta-Control stands out for its fully model-based nature, allowing rigorous analysis, generalizability, robustness, efficient parameter tuning, and reliable real-time execution.

Robot learning has proven to be a general and effective technique for programming manipulators. Imitation learning is able to teach robots solely from human demonstrations but is bottlenecked by the capabilities of the demonstrations. Reinforcement learning uses exploration to discover better behaviors; however, the space of possible improvements can be too large to start from scratch. And for both techniques, the learning difficulty increases proportional to the length of the manipulation task. Accounting for this, we propose SPIRE, a system that first uses Task and Motion Planning (TAMP) to decompose tasks into smaller learning subproblems and second combines imitation and reinforcement learning to maximize their strengths. We develop novel strategies to train learning agents when deployed in the context of a planning system. We evaluate SPIRE on a suite of long-horizon and contact-rich robot manipulation problems. We find that SPIRE outperforms prior approaches that integrate imitation learning, reinforcement learning, and planning by 35% to 50% in average task performance, is 6 times more data efficient in the number of human demonstrations needed to train proficient agents, and learns to complete tasks nearly twice as efficiently. View https://sites.google.com/view/spire-corl-2024 for more details.

Open-sourced, user-friendly tools form the bedrock of scientific advancement across disciplines. The widespread adoption of data-driven learning has led to remarkable progress in multi-fingered dexterity, bimanual manipulation, and applications ranging from logistics to home robotics. However, existing data collection platforms are often proprietary, costly, or tailored to specific robotic morphologies. We present OPEN TEACH, a new teleoperation system leveraging VR headsets to immerse users in mixed reality for intuitive robot control. built on the affordable Meta Quest 3, which costs $500, OPEN TEACH enables real-time control of various robots, including multi-fingered hands, bimanual arms, and mobile manipulators, through an easy-to-use app. Using natural hand gestures and movements, users can manipulate robots at up to 90Hz with smooth visual feedback and interface widgets offering closeup environment views. We demonstrate the versatility of OPEN TEACH across 38 tasks on different robots. A comprehensive user study indicates significant improvement in teleoperation capability over the AnyTeleop framework. Further experiments exhibit that the collected data is compatible with policy learning on 10 dexterous and contact-rich manipulation tasks. Currently supporting Franka, xArm, Jaco, Allegro, and Hello Stretch platforms, OPEN TEACH is fully open-sourced to promote broader adoption. Videos are available at https://anon-open-teach.github.io/.

Scaling up robot learning requires large and diverse datasets, and how to efficiently reuse collected data and transfer policies to new embodiments remains an open question. Emerging research such as the Open-X Embodiment (OXE) project has shown promise in leveraging skills by combining datasets including different robots. However, imbalances in the distribution of robot types and camera angles in many datasets make policies prone to overfit. To mitigate this issue, we propose RoVi-Aug, which leverages state-of-the-art image-to-image generative models to augment robot data by synthesizing demonstrations with different robots and camera views. Through extensive physical experiments, we show that, by training on robot- and viewpoint-augmented data, RoVi-Aug can zero-shot deploy on an unseen robot with significantly different camera angles. Compared to test-time adaptation algorithms such as Mirage, RoVi-Aug requires no extra processing at test time, does not assume known camera angles, and allows policy fine-tuning. Moreover, by co-training on both the original and augmented robot datasets, RoVi-Aug can learn multi-robot and multi-task policies, enabling more efficient transfer between robots and skills and improving success rates by up to 30%.

When we, humans, perform a task, we consider changes in environments such as objects' arrangement due to interactions with objects and other reasons; e.g., when we find a mug to clean, if it is already clean, we skip cleaning it. But even the state-of-the-art embodied agents often ignore changed environments when performing a task, leading to failure to complete the task, executing unnecessary actions, or fixing the mistake after it was made. Here, we propose Pre-emptive Action Revision by Environmental feeDback (PRED) that allows an embodied agent to revise their action in response to the perceived environmental status before it makes mistakes. We empirically validate PRED and observe that it outperforms the prior art on two challenging benchmarks in the virtual environment, TEACh and ALFRED, by noticeable margins in most metrics, including unseen success rates, with shorter execution time, implying an efficiently behaved agent. Furthermore, we demonstrate the effectiveness of the proposed method with real robot experiments.

Image-generation diffusion models have been fine-tuned to unlock new capabilities such as image-editing and novel view synthesis. Can we similarly unlock image-generation models for visuomotor control? We present GENIMA, a behavior-cloning agent that fine-tunes Stable Diffusion to “draw joint-actions” as targets on RGB images. These images are fed into a controller that maps the visual targets into a sequence of joint-positions. We study GENIMA on 25 RLBench and 9 real-world manipulation tasks. We find that, by lifting actions into image-space, internet pre-trained diffusion models can generate policies that outperform state- of-the-art visuomotor approaches, especially in robustness to scene perturbations and generalizing to novel objects. Our method is also competitive with 3D agents, despite lacking priors such as depth, keypoints, or motion-planners.

While grasp detection is an important part of any robotic manipulation pipeline, reliable and accurate grasp detection in $\\mathrm{SE}(3)$ remains a research challenge. Many robotics applications in unstructured environments such as the home or warehouse would benefit a lot from better grasp performance. This paper proposes a novel framework for detecting $\mathrm{SE}(3)$ grasp poses based on point cloud input. Our main contribution is to propose an $\mathrm{SE}(3)$-equivariant model that maps each point in the cloud to a continuous grasp quality function over the 2-sphere $S^2$ using a spherical harmonic basis. Compared with reasoning about a finite set of samples, this formulation improves the accuracy and efficiency of our model when a large number of samples would otherwise be needed. In order to accomplish this, we propose a novel variation on EquiFormerV2 that leverages a UNet-style backbone to enlarge the number of points the model can handle. Our resulting method, which we name OrbitGrasp, significantly outperforms baselines in both simulation and physical experiments.

We study the problem of mobile manipulation using legged robots equipped with an arm, namely legged loco-manipulation. The robot legs, while usually utilized for mobility, offer an opportunity to amplify the manipulation capabilities by conducting whole-body control. That is, the robot can control the legs and the arm at the same time to extend its workspace. We propose a framework that can conduct the whole-body control autonomously with visual observations. Our approach, namely Visual Whole-Body Control (VBC), is composed of a low-level policy using all degrees of freedom to track the body velocities along with the end-effector position, and a high-level policy proposing the velocities and end-effector position based on visual inputs. We train both levels of policies in simulation and perform Sim2Real transfer for real robot deployment. We perform extensive experiments and show significant improvements over baselines in picking up diverse objects in different configurations (heights, locations, orientations) and environments.

We present Im2Flow2Act, a scalable learning framework that enables robots to acquire real-world manipulation skills without the need of real-world robot training data. The key idea behind Im2Flow2Act is to use object flow as the manipulation interface, bridging domain gaps between different embodiments (i.e., human and robot) and training environments (i.e., real-world and simulated). Im2Flow2Act comprises two components: a flow generation network and a flow-conditioned policy. The flow generation network, trained on human demonstration videos, generates object flow from the initial scene image, conditioned on the task description. The flow-conditioned policy, trained on simulated robot play data, maps the generated object flow to robot actions to realize the desired object movements. By using flow as input, this policy can be directly deployed in the real world with a minimal sim-to-real gap. By leveraging real-world human videos and simulated robot play data, we bypass the challenges of teleoperating physical robots in the real world, resulting in a scalable system for diverse tasks. We demonstrate Im2Flow2Act's capabilities in a variety of real-world tasks, including the manipulation of rigid, articulated, and deformable objects.

Fast and accurate physics simulation is an essential component of robot learning, where robots can explore failure scenarios that are difficult to produce in the real world and learn from unlimited on-policy data. Yet, it remains challenging to incorporate RGB-color perception into the sim-to-real pipeline that matches the real world in its richness and realism. In this work, we train a robot dog in simulation for visual parkour. We propose a way to use generative models to synthesize diverse and physically accurate image sequences of the scene from the robot's ego-centric perspective. We present demonstrations of zero-shot transfer to the RGB-only observations of the real world on a robot equipped with a low-cost, off-the-shelf color camera.

Rapid advances in perception have enabled large pre-trained models to be used out of the box for transforming high-dimensional, noisy, and partial observations of the world into rich occupancy representations. However, the reliability of these models and consequently their safe integration onto robots remains unknown when deployed in environments unseen during training. In this work, we address this challenge by rigorously quantifying the uncertainty of pre-trained perception systems for object detection via a novel calibration technique based on conformal prediction. Crucially, this procedure guarantees robustness to distribution shifts in states when perceptual outputs are used in conjunction with a planner. As a result, the calibrated perception system can be used in combination with any safe planner to provide an end-to-end statistical assurance on safety in unseen environments. We evaluate the resulting approach, Perceive with Confidence (PwC), with experiments in simulation and on hardware where a quadruped robot navigates through previously unseen indoor, static environments. These experiments validate the safety assurances for obstacle avoidance provided by PwC and demonstrate up to 40% improvements in empirical safety compared to baselines.

Learning visuomotor policies for agile quadrotor flight presents significant difficulties, primarily from inefficient policy exploration caused by high-dimensional visual inputs and the need for precise and low-latency control. To address these challenges, we propose a novel approach that combines the performance of Reinforcement Learning (RL) and the sample efficiency of Imitation Learning (IL) in the task of vision-based autonomous drone racing. While RL provides a framework for learning high-performance controllers through trial and error, it faces challenges with sample efficiency and computational demands due to the high dimensionality of visual inputs. Conversely, IL efficiently learns from visual expert demonstrations, but it remains limited by the expert's performance and state distribution. To overcome these limitations, our policy learning framework integrates the strengths of both approaches. Our framework contains three phases: training a teacher policy using RL with privileged state information, distilling it into a student policy via IL, and adaptive fine-tuning via RL. Testing in both simulated and real-world scenarios shows our approach can not only learn in scenarios where RL from scratch fails but also outperforms existing IL methods in both robustness and performance, successfully navigating a quadrotor through a race course using only visual information.

Tactile and visual perception are both crucial for humans to perform fine-grained interactions with their environment. Developing similar multi-modal sensing capabilities for robots can significantly enhance and expand their manipulation skills. This paper introduces **3D-ViTac**, a multi-modal sensing and learning system designed for dexterous bimanual manipulation. Our system features tactile sensors equipped with dense sensing units, each covering an area of 3$mm^2$. These sensors are low-cost and flexible, providing detailed and extensive coverage of physical contacts, effectively complementing visual information. To integrate tactile and visual data, we fuse them into a unified 3D representation space that preserves their 3D structures and spatial relationships. The multi-modal representation can then be coupled with diffusion policies for imitation learning. Through concrete hardware experiments, we demonstrate that even low-cost robots can perform precise manipulations and significantly outperform vision-only policies, particularly in safe interactions with fragile items and executing long-horizon tasks involving in-hand manipulation. Our project page is available at https://binghao-huang.github.io/3D-ViTac/.

Ensuring the safety of autonomous vehicles (AVs) requires both accurate estimation of their performance and efficient discovery of potential failure cases. This paper introduces Bayesian adaptive multifidelity sampling (BAMS), which leverages the power of adaptive Bayesian sampling to achieve efficient discovery while simultaneously estimating the rate of adverse events. BAMS prioritizes exploration of regions with potentially low performance, leading to the identification of novel and critical scenarios that traditional methods might miss. Using real-world AV data we demonstrate that BAMS discovers 10 times as many issues as Monte Carlo (MC) and importance sampling (IS) baselines, while at the same time generating rate estimates with variances 15 and 6 times narrower than MC and IS baselines respectively.