Suneel Belkhale

Suneel Belkhale, Yuchen Cui, Dorsa Sadigh

In supervised learning, the question of data quality and curation has been over-shadowed in recent years by increasingly more powerful and expressive models that can ingest internet-scale data. However, in offline learning for robotics, we simply lack internet scale data, and so high quality datasets are a necessity. This is especially true in imitation learning (IL), a sample efficient paradigm for robot learning using expert demonstrations. Policies learned through IL suffer from state distribution shift at test time due to compounding errors in action prediction, which leads to unseen states that the policy cannot recover from. Instead of designing new algorithms to address distribution shift, an alternative perspective is to develop new ways of assessing and curating datasets. There is growing evidence that the same IL algorithms can have substantially different performance across different datasets. This calls for a formalism for defining metrics of "data quality" that can further be leveraged for data curation. In this work, we take the first step toward formalizing data quality for imitation learning through the lens of distribution shift: a high quality dataset encourages the policy to stay in distribution at test time. We propose two fundamental properties that shape the quality of a dataset: i) action divergence: the mismatch between the expert and learned policy at certain states; and ii) transition diversity: the noise present in the system for a given state and action. We investigate the combined effect of these two key properties in imitation learning theoretically, and we empirically analyze models trained on a variety of different data sources. We show that state diversity is not always beneficial, and we demonstrate how action divergence and transition diversity interact in practice.

Jennifer Grannen, Yilin Wu, Suneel Belkhale et al.

A robotic feeding system must be able to acquire a variety of foods. Prior bite acquisition works consider single-arm spoon scooping or fork skewering, which do not generalize to foods with complex geometries and deformabilities. For example, when acquiring a group of peas, skewering could smoosh the peas while scooping without a barrier could result in chasing the peas on the plate. In order to acquire foods with such diverse properties, we propose stabilizing food items during scooping using a second arm, for example, by pushing peas against the spoon with a flat surface to prevent dispersion. The added stabilizing arm can lead to new challenges. Critically, this arm should stabilize the food scene without interfering with the acquisition motion, which is especially difficult for easily breakable high-risk food items like tofu. These high-risk foods can break between the pusher and spoon during scooping, which can lead to food waste falling out of the spoon. We propose a general bimanual scooping primitive and an adaptive stabilization strategy that enables successful acquisition of a diverse set of food geometries and physical properties. Our approach, CARBS: Coordinated Acquisition with Reactive Bimanual Scooping, learns to stabilize without impeding task progress by identifying high-risk foods and robustly scooping them using closed-loop visual feedback. We find that CARBS is able to generalize across food shape, size, and deformability and is additionally able to manipulate multiple food items simultaneously. CARBS achieves 87.0% success on scooping rigid foods, which is 25.8% more successful than a single-arm baseline, and reduces food breakage by 16.2% compared to an analytical baseline. Videos can be found at https://sites.google.com/view/bimanualscoop-corl22/home .

Priya Sundaresan, Suneel Belkhale, Dorsa Sadigh

Acquiring food items with a fork poses an immense challenge to a robot-assisted feeding system, due to the wide range of material properties and visual appearances present across food groups. Deformable foods necessitate different skewering strategies than firm ones, but inferring such characteristics for several previously unseen items on a plate remains nontrivial. Our key insight is to leverage visual and haptic observations during interaction with an item to rapidly and reactively plan skewering motions. We learn a generalizable, multimodal representation for a food item from raw sensory inputs which informs the optimal skewering strategy. Given this representation, we propose a zero-shot framework to sense visuo-haptic properties of a previously unseen item and reactively skewer it, all within a single interaction. Real-robot experiments with foods of varying levels of visual and textural diversity demonstrate that our multimodal policy outperforms baselines which do not exploit both visual and haptic cues or do not reactively plan. Across 6 plates of different food items, our proposed framework achieves 71% success over 69 skewering attempts total. Supplementary material, datasets, code, and videos are available on our website: https://sites.google.com/view/hapticvisualnet-corl22/home

Priya Sundaresan, Suneel Belkhale, Dorsa Sadigh et al.

While natural language offers a convenient shared interface for humans and robots, enabling robots to interpret and follow language commands remains a longstanding challenge in manipulation. A crucial step to realizing a performant instruction-following robot is achieving semantic manipulation, where a robot interprets language at different specificities, from high-level instructions like "Pick up the stuffed animal" to more detailed inputs like "Grab the left ear of the elephant." To tackle this, we propose Keypoints + Instructions to Execution (KITE), a two-step framework for semantic manipulation which attends to both scene semantics (distinguishing between different objects in a visual scene) and object semantics (precisely localizing different parts within an object instance). KITE first grounds an input instruction in a visual scene through 2D image keypoints, providing a highly accurate object-centric bias for downstream action inference. Provided an RGB-D scene observation, KITE then executes a learned keypoint-conditioned skill to carry out the instruction. The combined precision of keypoints and parameterized skills enables fine-grained manipulation with generalization to scene and object variations. Empirically, we demonstrate KITE in 3 real-world environments: long-horizon 6-DoF tabletop manipulation, semantic grasping, and a high-precision coffee-making task. In these settings, KITE achieves a 75%, 70%, and 71% overall success rate for instruction-following, respectively. KITE outperforms frameworks that opt for pre-trained visual language models over keypoint-based grounding, or omit skills in favor of end-to-end visuomotor control, all while being trained from fewer or comparable amounts of demonstrations. Supplementary material, datasets, code, and videos can be found on our website: http://tinyurl.com/kite-site.

Suneel Belkhale, Tianli Ding, Ted Xiao et al.

Language provides a way to break down complex concepts into digestible pieces. Recent works in robot imitation learning use language-conditioned policies that predict actions given visual observations and the high-level task specified in language. These methods leverage the structure of natural language to share data between semantically similar tasks (e.g., "pick coke can" and "pick an apple") in multi-task datasets. However, as tasks become more semantically diverse (e.g., "pick coke can" and "pour cup"), sharing data between tasks becomes harder, so learning to map high-level tasks to actions requires much more demonstration data. To bridge tasks and actions, our insight is to teach the robot the language of actions, describing low-level motions with more fine-grained phrases like "move arm forward". Predicting these language motions as an intermediate step between tasks and actions forces the policy to learn the shared structure of low-level motions across seemingly disparate tasks. Furthermore, a policy that is conditioned on language motions can easily be corrected during execution through human-specified language motions. This enables a new paradigm for flexible policies that can learn from human intervention in language. Our method RT-H builds an action hierarchy using language motions: it first learns to predict language motions, and conditioned on this and the high-level task, it predicts actions, using visual context at all stages. We show that RT-H leverages this language-action hierarchy to learn policies that are more robust and flexible by effectively tapping into multi-task datasets. We show that these policies not only allow for responding to language interventions, but can also learn from such interventions and outperform methods that learn from teleoperated interventions. Our website and videos are found at https://rt-hierarchy.github.io.

Katie Kang, Suneel Belkhale, Gregory Kahn et al.

Deep reinforcement learning provides a promising approach for vision-based control of real-world robots. However, the generalization of such models depends critically on the quantity and variety of data available for training. This data can be difficult to obtain for some types of robotic systems, such as fragile, small-scale quadrotors. Simulated rendering and physics can provide for much larger datasets, but such data is inherently of lower quality: many of the phenomena that make the real-world autonomous flight problem challenging, such as complex physics and air currents, are modeled poorly or not at all, and the systematic differences between simulation and the real world are typically impossible to eliminate. In this work, we investigate how data from both simulation and the real world can be combined in a hybrid deep reinforcement learning algorithm. Our method uses real-world data to learn about the dynamics of the system, and simulated data to learn a generalizable perception system that can enable the robot to avoid collisions using only a monocular camera. We demonstrate our approach on a real-world nano aerial vehicle collision avoidance task, showing that with only an hour of real-world data, the quadrotor can avoid collisions in new environments with various lighting conditions and geometry. Code, instructions for building the aerial vehicles, and videos of the experiments can be found at github.com/gkahn13/GtS

Jensen Gao, Suneel Belkhale, Sudeep Dasari et al.

Machine learning for robotics promises to unlock generalization to novel tasks and environments. Guided by this promise, many recent works have focused on scaling up robot data collection and developing larger, more expressive policies to achieve this. But how do we measure progress towards this goal of policy generalization in practice? Evaluating and quantifying generalization is the Wild West of modern robotics, with each work proposing and measuring different types of generalization in their own, often difficult to reproduce, settings. In this work, our goal is (1) to outline the forms of generalization we believe are important in robot manipulation in a comprehensive and fine-grained manner, and (2) to provide reproducible guidelines for measuring these notions of generalization. We first propose STAR-Gen, a taxonomy of generalization for robot manipulation structured around visual, semantic, and behavioral generalization. We discuss how our taxonomy encompasses most prior notions of generalization in robotics. Next, we instantiate STAR-Gen with a concrete real-world benchmark based on the widely-used Bridge V2 dataset. We evaluate a variety of state-of-the-art models on this benchmark to demonstrate the utility of our taxonomy in practice. Our taxonomy of generalization can yield many interesting insights into existing models: for example, we observe that current vision-language-action models struggle with various types of semantic generalization, despite the promise of pre-training on internet-scale language datasets. We believe STAR-Gen and our guidelines can improve the dissemination and evaluation of progress towards generalization in robotics, which we hope will guide model design and future data collection efforts. We provide videos and demos at our website stargen-taxonomy.github.io.

Jaden Clark, Suvir Mirchandani, Dorsa Sadigh et al. · stanford

End-to-end imitation learning offers a promising approach for training robot policies. However, generalizing to new settings remains a significant challenge. Although large-scale robot demonstration datasets have shown potential for inducing generalization, they are resource-intensive to scale. In contrast, human video data is abundant and diverse, presenting an attractive alternative. Yet, these human-video datasets lack action labels, complicating their use in imitation learning. Existing methods attempt to extract grounded action representations (e.g., hand poses), but resulting policies struggle to bridge the embodiment gap between human and robot actions. We propose an alternative approach: leveraging language-based reasoning from human videos-essential for guiding robot actions-to train generalizable robot policies. Building on recent advances in reasoning-based policy architectures, we introduce Reasoning through Action-free Data (RAD). RAD learns from both robot demonstration data (with reasoning and action labels) and action-free human video data (with only reasoning labels). The robot data teaches the model to map reasoning to low-level actions, while the action-free data enhances reasoning capabilities. Additionally, we will release a new dataset of 3,377 human-hand demonstrations with reasoning annotations compatible with the Bridge V2 benchmark and aimed at facilitating future research on reasoning-driven robot learning. Our experiments show that RAD enables effective transfer across the embodiment gap, allowing robots to perform tasks seen only in action-free data. Furthermore, scaling up action-free reasoning data significantly improves policy performance and generalization to novel tasks. These results highlight the promise of reasoning-driven learning from action-free datasets for advancing generalizable robot control. Project page: https://rad-generalization.github.io

Suvir Mirchandani, Suneel Belkhale, Joey Hejna et al. · stanford

A long-standing goal in robot learning is to develop methods for robots to acquire new skills autonomously. While reinforcement learning (RL) comes with the promise of enabling autonomous data collection, it remains challenging to scale in the real-world partly due to the significant effort required for environment design and instrumentation, including the need for designing reset functions or accurate success detectors. On the other hand, imitation learning (IL) methods require little to no environment design effort, but instead require significant human supervision in the form of collected demonstrations. To address these shortcomings, recent works in autonomous IL start with an initial seed dataset of human demonstrations that an autonomous policy can bootstrap from. While autonomous IL approaches come with the promise of addressing the challenges of autonomous RL as well as pure IL strategies, in this work, we posit that such techniques do not deliver on this promise and are still unable to scale up autonomous data collection in the real world. Through a series of real-world experiments, we demonstrate that these approaches, when scaled up to realistic settings, face much of the same scaling challenges as prior attempts in RL in terms of environment design. Further, we perform a rigorous study of autonomous IL methods across different data scales and 7 simulation and real-world tasks, and demonstrate that while autonomous data collection can modestly improve performance, simply collecting more human data often provides significantly more improvement. Our work suggests a negative result: that scaling up autonomous data collection for learning robot policies for real-world tasks is more challenging and impractical than what is suggested in prior work. We hope these insights about the core challenges of scaling up data collection help inform future efforts in autonomous learning.

Andy Shih, Suneel Belkhale, Stefano Ermon et al.

Diffusion models are powerful generative models but suffer from slow sampling, often taking 1000 sequential denoising steps for one sample. As a result, considerable efforts have been directed toward reducing the number of denoising steps, but these methods hurt sample quality. Instead of reducing the number of denoising steps (trading quality for speed), in this paper we explore an orthogonal approach: can we run the denoising steps in parallel (trading compute for speed)? In spite of the sequential nature of the denoising steps, we show that surprisingly it is possible to parallelize sampling via Picard iterations, by guessing the solution of future denoising steps and iteratively refining until convergence. With this insight, we present ParaDiGMS, a novel method to accelerate the sampling of pretrained diffusion models by denoising multiple steps in parallel. ParaDiGMS is the first diffusion sampling method that enables trading compute for speed and is even compatible with existing fast sampling techniques such as DDIM and DPMSolver. Using ParaDiGMS, we improve sampling speed by 2-4x across a range of robotics and image generation models, giving state-of-the-art sampling speeds of 0.2s on 100-step DiffusionPolicy and 14.6s on 1000-step StableDiffusion-v2 with no measurable degradation of task reward, FID score, or CLIP score.

Suneel Belkhale, Ethan K. Gordon, Yuxiao Chen et al.

Robot-assisted feeding in household environments is challenging because it requires robots to generate trajectories that effectively bring food items of varying shapes and sizes into the mouth while making sure the user is comfortable. Our key insight is that in order to solve this challenge, robots must balance the efficiency of feeding a food item with the comfort of each individual bite. We formalize comfort and efficiency as heuristics to incorporate in motion planning. We present an approach based on heuristics-guided bi-directional Rapidly-exploring Random Trees (h-BiRRT) that selects bite transfer trajectories of arbitrary food item geometries and shapes using our developed bite efficiency and comfort heuristics and a learned constraint model. Real-robot evaluations show that optimizing both comfort and efficiency significantly outperforms a fixed-pose based method, and users preferred our method significantly more than that of a method that maximizes only user comfort. Videos and Appendices are found on our website: https://sites.google.com/view/comfortbitetransfer-icra22/home.

Suneel Belkhale, Rachel Li, Gregory Kahn et al.

Transporting suspended payloads is challenging for autonomous aerial vehicles because the payload can cause significant and unpredictable changes to the robot's dynamics. These changes can lead to suboptimal flight performance or even catastrophic failure. Although adaptive control and learning-based methods can in principle adapt to changes in these hybrid robot-payload systems, rapid mid-flight adaptation to payloads that have a priori unknown physical properties remains an open problem. We propose a meta-learning approach that "learns how to learn" models of altered dynamics within seconds of post-connection flight data. Our experiments demonstrate that our online adaptation approach outperforms non-adaptive methods on a series of challenging suspended payload transportation tasks. Videos and other supplemental material are available on our website: https://sites.google.com/view/meta-rl-for-flight