Zackory Erickson

Jeremy Tien, Jerry Zhi-Yang He, Zackory Erickson et al. · cmu

Learning policies via preference-based reward learning is an increasingly popular method for customizing agent behavior, but has been shown anecdotally to be prone to spurious correlations and reward hacking behaviors. While much prior work focuses on causal confusion in reinforcement learning and behavioral cloning, we focus on a systematic study of causal confusion and reward misidentification when learning from preferences. In particular, we perform a series of sensitivity and ablation analyses on several benchmark domains where rewards learned from preferences achieve minimal test error but fail to generalize to out-of-distribution states -- resulting in poor policy performance when optimized. We find that the presence of non-causal distractor features, noise in the stated preferences, and partial state observability can all exacerbate reward misidentification. We also identify a set of methods with which to interpret misidentified learned rewards. In general, we observe that optimizing misidentified rewards drives the policy off the reward's training distribution, resulting in high predicted (learned) rewards but low true rewards. These findings illuminate the susceptibility of preference learning to reward misidentification and causal confusion -- failure to consider even one of many factors can result in unexpected, undesirable behavior.

Jerry Zhi-Yang He, Aditi Raghunathan, Daniel S. Brown et al. · cmu

Recent work in sim2real has successfully enabled robots to act in physical environments by training in simulation with a diverse ''population'' of environments (i.e. domain randomization). In this work, we focus on enabling generalization in assistive tasks: tasks in which the robot is acting to assist a user (e.g. helping someone with motor impairments with bathing or with scratching an itch). Such tasks are particularly interesting relative to prior sim2real successes because the environment now contains a human who is also acting. This complicates the problem because the diversity of human users (instead of merely physical environment parameters) is more difficult to capture in a population, thus increasing the likelihood of encountering out-of-distribution (OOD) human policies at test time. We advocate that generalization to such OOD policies benefits from (1) learning a good latent representation for human policies that test-time humans can accurately be mapped to, and (2) making that representation adaptable with test-time interaction data, instead of relying on it to perfectly capture the space of human policies based on the simulated population only. We study how to best learn such a representation by evaluating on purposefully constructed OOD test policies. We find that sim2real methods that encode environment (or population) parameters and work well in tasks that robots do in isolation, do not work well in assistance. In assistance, it seems crucial to train the representation based on the history of interaction directly, because that is what the robot will have access to at test time. Further, training these representations to then predict human actions not only gives them better structure, but also enables them to be fine-tuned at test-time, when the robot observes the partner act. https://adaptive-caregiver.github.io.

Yufei Wang, Zhou Xian, Feng Chen et al. · cmu

We present RoboGen, a generative robotic agent that automatically learns diverse robotic skills at scale via generative simulation. RoboGen leverages the latest advancements in foundation and generative models. Instead of directly using or adapting these models to produce policies or low-level actions, we advocate for a generative scheme, which uses these models to automatically generate diversified tasks, scenes, and training supervisions, thereby scaling up robotic skill learning with minimal human supervision. Our approach equips a robotic agent with a self-guided propose-generate-learn cycle: the agent first proposes interesting tasks and skills to develop, and then generates corresponding simulation environments by populating pertinent objects and assets with proper spatial configurations. Afterwards, the agent decomposes the proposed high-level task into sub-tasks, selects the optimal learning approach (reinforcement learning, motion planning, or trajectory optimization), generates required training supervision, and then learns policies to acquire the proposed skill. Our work attempts to extract the extensive and versatile knowledge embedded in large-scale models and transfer them to the field of robotics. Our fully generative pipeline can be queried repeatedly, producing an endless stream of skill demonstrations associated with diverse tasks and environments.

Akhil Padmanabha, Sonal Choudhary, Carmel Majidi et al. · cmu

An objective measurement of chronic itch is necessary for improvements in patient care for numerous medical conditions. While wearables have shown promise for scratch detection, they are currently unable to estimate scratch intensity, preventing a comprehensive understanding of the effect of itch on an individual. In this work, we present a framework for the estimation of scratch intensity in addition to the detection of scratch. This is accomplished with a multimodal ring device, consisting of an accelerometer and a contact microphone, a pressure-sensitive tablet for capturing ground truth intensity values, and machine learning algorithms for regression of scratch intensity on a 0-600 milliwatts (mW) power scale that can be mapped to a 0-10 continuous scale. We evaluate the performance of our algorithms on 20 individuals using leave one subject out cross-validation and using data from 14 additional participants, we show that our algorithms achieve clinically-relevant discrimination of scratching intensity levels. By doing so, our device enables the quantification of the substantial variations in the interpretation of the 0-10 scale frequently utilized in patient self-reported clinical assessments. This work demonstrates that a finger-worn device can provide multidimensional, objective, real-time measures for the action of scratching.

Alberta Longhini, Marco Moletta, Alfredo Reichlin et al. · cmu

We study the problem of learning graph dynamics of deformable objects that generalizes to unknown physical properties. Our key insight is to leverage a latent representation of elastic physical properties of cloth-like deformable objects that can be extracted, for example, from a pulling interaction. In this paper we propose EDO-Net (Elastic Deformable Object - Net), a model of graph dynamics trained on a large variety of samples with different elastic properties that does not rely on ground-truth labels of the properties. EDO-Net jointly learns an adaptation module, and a forward-dynamics module. The former is responsible for extracting a latent representation of the physical properties of the object, while the latter leverages the latent representation to predict future states of cloth-like objects represented as graphs. We evaluate EDO-Net both in simulation and real world, assessing its capabilities of: 1) generalizing to unknown physical properties, 2) transferring the learned representation to new downstream tasks.

Yufei Wang, David Held, Zackory Erickson · cmu

Robotic manipulation of highly deformable cloth presents a promising opportunity to assist people with several daily tasks, such as washing dishes; folding laundry; or dressing, bathing, and hygiene assistance for individuals with severe motor impairments. In this work, we introduce a formulation that enables a collaborative robot to perform visual haptic reasoning with cloth -- the act of inferring the location and magnitude of applied forces during physical interaction. We present two distinct model representations, trained in physics simulation, that enable haptic reasoning using only visual and robot kinematic observations. We conducted quantitative evaluations of these models in simulation for robot-assisted dressing, bathing, and dish washing tasks, and demonstrate that the trained models can generalize across different tasks with varying interactions, human body sizes, and object shapes. We also present results with a real-world mobile manipulator, which used our simulation-trained models to estimate applied contact forces while performing physically assistive tasks with cloth. Videos can be found at our project webpage.

Jerry Zhi-Yang He, Zackory Erickson, Daniel S. Brown et al. · cmu

Our ultimate goal is to build robust policies for robots that assist people. What makes this hard is that people can behave unexpectedly at test time, potentially interacting with the robot outside its training distribution and leading to failures. Even just measuring robustness is a challenge. Adversarial perturbations are the default, but they can paint the wrong picture: they can correspond to human motions that are unlikely to occur during natural interactions with people. A robot policy might fail under small adversarial perturbations but work under large natural perturbations. We propose that capturing robustness in these interactive settings requires constructing and analyzing the entire natural-adversarial frontier: the Pareto-frontier of human policies that are the best trade-offs between naturalness and low robot performance. We introduce RIGID, a method for constructing this frontier by training adversarial human policies that trade off between minimizing robot reward and acting human-like (as measured by a discriminator). On an Assistive Gym task, we use RIGID to analyze the performance of standard collaborative Reinforcement Learning, as well as the performance of existing methods meant to increase robustness. We also compare the frontier RIGID identifies with the failures identified in expert adversarial interaction, and with naturally-occurring failures during user interaction. Overall, we find evidence that RIGID can provide a meaningful measure of robustness predictive of deployment performance, and uncover failure cases in human-robot interaction that are difficult to find manually. https://ood-human.github.io.

Janavi Gupta, Kavya Puthuveetil, Dimitra Tsakona et al.

Assistive robotic systems have shown growing potential to improve the quality of life of those with disabilities. As researchers explore the automation of various caregiving tasks, considerations for how the technology can still preserve the user's sense of control become paramount to ensuring that robotic systems are aligned with fundamental user needs and motivations. In this work, we present two previously developed systems as design cases through which to explore an interaction paradigm that we call implicit control, where the behavior of an autonomous robot is modified based on users' natural behavioral cues, instead of some direct input. Our selected design cases, unlike systems in past work, specifically probe users' perception of the interaction. We find, from a new thematic analysis of qualitative feedback on both cases, that designing for effective implicit control enables both a reduction in perceived workload and the preservation of the users' sense of control through the system's intuitiveness and responsiveness, contextual awareness, and ability to adapt to preferences. We further derive a set of core guidelines for designers in deciding when and how to apply implicit interaction paradigms for their assistive applications.

Hongyi Chen, Tony Dong, Tiancheng Wu et al.

Multi-finger robotic hand manipulation and grasping are challenging due to the high-dimensional action space and the difficulty of acquiring large-scale training data. Existing approaches largely rely on human teleoperation with wearable devices or specialized sensing equipment to capture hand-object interactions, which limits scalability. In this work, we propose VIDEOMANIP, a device-free framework that learns dexterous manipulation directly from RGB human videos. Leveraging recent advances in computer vision, VIDEOMANIP reconstructs explicit 4D robot-object trajectories from monocular videos by estimating human hand poses, object meshes, and retargets the reconstructed human motions to robotic hands for manipulation learning. To make the reconstructed robot data suitable for dexterous manipulation training, we introduce hand-object contact optimization with interaction-centric grasp modeling, as well as a demonstration synthesis strategy that generates diverse training trajectories from a single video, enabling generalizable policy learning without additional robot demonstrations. In simulation, the learned grasping model achieves a 70.25% success rate across 20 diverse objects using the Inspire Hand. In the real world, manipulation policies trained from RGB videos achieve an average 62.86% success rate across seven tasks using the LEAP Hand, outperforming retargeting-based methods by 15.87%. Project videos are available at videomanip.github.io.

Weikang Wan, Ziyu Wang, Yufei Wang et al. · cmu

This paper introduces DiffTORI, which utilizes Differentiable Trajectory Optimization as the policy representation to generate actions for deep Reinforcement and Imitation learning. Trajectory optimization is a powerful and widely used algorithm in control, parameterized by a cost and a dynamics function. The key to our approach is to leverage the recent progress in differentiable trajectory optimization, which enables computing the gradients of the loss with respect to the parameters of trajectory optimization. As a result, the cost and dynamics functions of trajectory optimization can be learned end-to-end. DiffTORI addresses the ``objective mismatch'' issue of prior model-based RL algorithms, as the dynamics model in DiffTORI is learned to directly maximize task performance by differentiating the policy gradient loss through the trajectory optimization process. We further benchmark DiffTORI for imitation learning on standard robotic manipulation task suites with high-dimensional sensory observations and compare our method to feed-forward policy classes as well as Energy-Based Models (EBM) and Diffusion. Across 15 model-based RL tasks and 35 imitation learning tasks with high-dimensional image and point cloud inputs, DiffTORI outperforms prior state-of-the-art methods in both domains. Our code is available at https://github.com/wkwan7/DiffTORI.

Jehan Yang, Maxwell Soh, Vivianna Lieu et al. · cmu

This paper introduces the first generalization and adaptation benchmark using machine learning for evaluating out-of-distribution performance of electromyography (EMG) classification algorithms. The ability of an EMG classifier to handle inputs drawn from a different distribution than the training distribution is critical for real-world deployment as a control interface. By predicting the user's intended gesture using EMG signals, we can create a wearable solution to control assistive technologies, such as computers, prosthetics, and mobile manipulator robots. This new out-of-distribution benchmark consists of two major tasks that have utility for building robust and adaptable control interfaces: 1) intersubject classification and 2) adaptation using train-test splits for time-series. This benchmark spans nine datasets--the largest collection of EMG datasets in a benchmark. Among these, a new dataset is introduced, featuring a novel, easy-to-wear high-density EMG wearable for data collection. The lack of open-source benchmarks has made comparing accuracy results between papers challenging for the EMG research community. This new benchmark provides researchers with a valuable resource for analyzing practical measures of out-of-distribution performance for EMG datasets. Our code and data from our new dataset can be found at emgbench.github.io.

Kavya Puthuveetil, Charles C. Kemp, Zackory Erickson

While robots present an opportunity to provide physical assistance to older adults and people with mobility impairments in bed, people frequently rest in bed with blankets that cover the majority of their body. To provide assistance for many daily self-care tasks, such as bathing, dressing, or ambulating, a caregiver must first uncover blankets from part of a person's body. In this work, we introduce a formulation for robotic bedding manipulation around people in which a robot uncovers a blanket from a target body part while ensuring the rest of the human body remains covered. We compare two approaches for optimizing policies which provide a robot with grasp and release points that uncover a target part of the body: 1) reinforcement learning and 2) self-supervised learning with optimization to generate training data. We trained and conducted evaluations of these policies in physics simulation environments that consist of a deformable cloth mesh covering a simulated human lying supine on a bed. In addition, we transfer simulation-trained policies to a real mobile manipulator and demonstrate that it can uncover a blanket from target body parts of a manikin lying in bed. Source code is available online.

Zackory Erickson, Vamsee Gangaram, Ariel Kapusta et al.

Autonomous robots have the potential to serve as versatile caregivers that improve quality of life for millions of people worldwide. Yet, conducting research in this area presents numerous challenges, including the risks of physical interaction between people and robots. Physics simulations have been used to optimize and train robots for physical assistance, but have typically focused on a single task. In this paper, we present Assistive Gym, an open source physics simulation framework for assistive robots that models multiple tasks. It includes six simulated environments in which a robotic manipulator can attempt to assist a person with activities of daily living (ADLs): itch scratching, drinking, feeding, body manipulation, dressing, and bathing. Assistive Gym models a person's physical capabilities and preferences for assistance, which are used to provide a reward function. We present baseline policies trained using reinforcement learning for four different commercial robots in the six environments. We demonstrate that modeling human motion results in better assistance and we compare the performance of different robots. Overall, we show that Assistive Gym is a promising tool for assistive robotics research.

Yufei Wang, Zhanyi Sun, Jesse Zhang et al. · cmu

Reward engineering has long been a challenge in Reinforcement Learning (RL) research, as it often requires extensive human effort and iterative processes of trial-and-error to design effective reward functions. In this paper, we propose RL-VLM-F, a method that automatically generates reward functions for agents to learn new tasks, using only a text description of the task goal and the agent's visual observations, by leveraging feedbacks from vision language foundation models (VLMs). The key to our approach is to query these models to give preferences over pairs of the agent's image observations based on the text description of the task goal, and then learn a reward function from the preference labels, rather than directly prompting these models to output a raw reward score, which can be noisy and inconsistent. We demonstrate that RL-VLM-F successfully produces effective rewards and policies across various domains - including classic control, as well as manipulation of rigid, articulated, and deformable objects - without the need for human supervision, outperforming prior methods that use large pretrained models for reward generation under the same assumptions. Videos can be found on our project website: https://rlvlmf2024.github.io/

Junxiang Wang, Xinwen Xu, Tiancheng Wu et al.

Developing autonomous physical human-robot interaction (pHRI) systems is limited by the scarcity of large-scale training data to learn robust robot behaviors for real-world applications. In this paper, we introduce a zero-shot "text2sim2real" generative simulation framework that automatically synthesizes diverse pHRI scenarios from high-level natural-language prompts. Leveraging Large Language Models (LLMs) and Vision-Language Models (VLMs), our pipeline procedurally generates soft-body human models, scene layouts, and robot motion trajectories for assistive tasks. We utilize this framework to autonomously collect large-scale synthetic demonstration datasets and then train vision-based imitation learning policies operating on segmented point clouds. We evaluate our approach through a user study on two physically assistive tasks: scratching and bathing. Our learned policies successfully achieve zero-shot sim-to-real transfer, attaining success rates exceeding 80% and demonstrating resilience to unscripted human motion. Overall, we introduce the first generative simulation pipeline for pHRI applications, automating simulation environment synthesis, data collection, and policy learning. Additional information may be found on our project website: https://rchi-lab.github.io/gen_phri/

Akhil Padmanabha, Jessie Yuan, Janavi Gupta et al. · cmu

Physically assistive robots present an opportunity to significantly increase the well-being and independence of individuals with motor impairments or other forms of disability who are unable to complete activities of daily living. Speech interfaces, especially ones that utilize Large Language Models (LLMs), can enable individuals to effectively and naturally communicate high-level commands and nuanced preferences to robots. Frameworks for integrating LLMs as interfaces to robots for high level task planning and code generation have been proposed, but fail to incorporate human-centric considerations which are essential while developing assistive interfaces. In this work, we present a framework for incorporating LLMs as speech interfaces for physically assistive robots, constructed iteratively with 3 stages of testing involving a feeding robot, culminating in an evaluation with 11 older adults at an independent living facility. We use both quantitative and qualitative data from the final study to validate our framework and additionally provide design guidelines for using LLMs as speech interfaces for assistive robots. Videos and supporting files are located on our project website: https://sites.google.com/andrew.cmu.edu/voicepilot/

Sreyas Venkataraman, Yufei Wang, Ziyu Wang et al. · cmu

Offline reinforcement learning can enable policy learning from pre-collected, sub-optimal datasets without online interactions. This makes it ideal for real-world robots and safety-critical scenarios, where collecting online data or expert demonstrations is slow, costly, and risky. However, most existing offline RL works assume the dataset is already labeled with the task rewards, a process that often requires significant human effort, especially when ground-truth states are hard to ascertain (e.g., in the real-world). In this paper, we build on prior work, specifically RL-VLM-F, and propose a novel system that automatically generates reward labels for offline datasets using preference feedback from a vision-language model and a text description of the task. Our method then learns a policy using offline RL with the reward-labeled dataset. We demonstrate the system's applicability to a complex real-world robot-assisted dressing task, where we first learn a reward function using a vision-language model on a sub-optimal offline dataset, and then we use the learned reward to employ Implicit Q learning to develop an effective dressing policy. Our method also performs well in simulation tasks involving the manipulation of rigid and deformable objects, and significantly outperform baselines such as behavior cloning and inverse RL. In summary, we propose a new system that enables automatic reward labeling and policy learning from unlabeled, sub-optimal offline datasets.

Abhishek Tandon, Anujraaj Goyal, Henry M. Clever et al. · cmu

Accurately predicting the 3D human posture and the pressure exerted on the body for people resting in bed, visualized as a body mesh (3D pose & shape) with a 3D pressure map, holds significant promise for healthcare applications, particularly, in the prevention of pressure ulcers. Current methods focus on singular facets of the problem -- predicting only 2D/3D poses, generating 2D pressure images, predicting pressure only for certain body regions instead of the full body, or forming indirect approximations to the 3D pressure map. In contrast, we introduce BodyMAP, which jointly predicts the human body mesh and 3D applied pressure map across the entire human body. Our network leverages multiple visual modalities, incorporating both a depth image of a person in bed and its corresponding 2D pressure image acquired from a pressure-sensing mattress. The 3D pressure map is represented as a pressure value at each mesh vertex and thus allows for precise localization of high-pressure regions on the body. Additionally, we present BodyMAP-WS, a new formulation of pressure prediction in which we implicitly learn pressure in 3D by aligning sensed 2D pressure images with a differentiable 2D projection of the predicted 3D pressure maps. In evaluations with real-world human data, our method outperforms the current state-of-the-art technique by 25% on both body mesh and 3D applied pressure map prediction tasks for people in bed.

Jessie Yuan, Janavi Gupta, Akhil Padmanabha et al. · cmu

Physically assistive robots present an opportunity to significantly increase the well-being and independence of individuals with motor impairments or other forms of disability who are unable to complete activities of daily living (ADLs). Speech interfaces, especially ones that utilize Large Language Models (LLMs), can enable individuals to effectively and naturally communicate high-level commands and nuanced preferences to robots. In this work, we demonstrate an LLM-based speech interface for a commercially available assistive feeding robot. Our system is based on an iteratively designed framework, from the paper "VoicePilot: Harnessing LLMs as Speech Interfaces for Physically Assistive Robots," that incorporates human-centric elements for integrating LLMs as interfaces for robots. It has been evaluated through a user study with 11 older adults at an independent living facility. Videos are located on our project website: https://sites.google.com/andrew.cmu.edu/voicepilot/.

Zheyuan Hu, Robyn Wu, Naveen Enock et al.

Modern paradigms for robot imitation train expressive policy architectures on large amounts of human demonstration data. Yet performance on contact-rich, deformable-object, and long-horizon tasks plateau far below perfect execution, even with thousands of expert demonstrations. This is due to the inefficiency of existing ``expert'' data collection procedures based on human teleoperation. To address this issue, we introduce RaC, a new phase of training on human-in-the-loop rollouts after imitation learning pre-training. In RaC, we fine-tune a robotic policy on human intervention trajectories that illustrate recovery and correction behaviors. Specifically, during a policy rollout, human operators intervene when failure appears imminent, first rewinding the robot back to a familiar, in-distribution state and then providing a corrective segment that completes the current sub-task. Training on this data composition expands the robotic skill repertoire to include retry and adaptation behaviors, which we show are crucial for boosting both efficiency and robustness on long-horizon tasks. Across three real-world bimanual control tasks: shirt hanging, airtight container lid sealing, takeout box packing, and a simulated assembly task, RaC outperforms the prior state-of-the-art using 10$\times$ less data collection time and samples. We also show that RaC enables test-time scaling: the performance of the trained RaC policy scales linearly in the number of recovery maneuvers it exhibits. Videos of the learned policy are available at https://rac-scaling-robot.github.io/.

Yiran Tao, Jehan Yang, Dan Ding et al. · cmu

Teleoperating high degrees-of-freedom (DoF) robotic manipulators via low-DoF controllers like joysticks often requires frequent switching between control modes, where each mode maps controller movements to specific robot actions. Manually performing this frequent switching can make teleoperation cumbersome and inefficient. On the other hand, existing automatic mode-switching solutions, such as heuristic-based or learning-based methods, are often task-specific and lack generalizability. In this paper, we introduce LLM-Driven Automatic Mode Switching (LAMS), a novel approach that leverages Large Language Models (LLMs) to automatically switch control modes based on task context. Unlike existing methods, LAMS requires no prior task demonstrations and incrementally improves by integrating user-generated mode-switching examples. We validate LAMS through an ablation study and a user study with 10 participants on complex, long-horizon tasks, demonstrating that LAMS effectively reduces manual mode switches, is preferred over alternative methods, and improves performance over time. The project website with supplementary materials is at https://lams-assistance.github.io/.

Hongyi Chen, Yunchao Yao, Yufei Ye et al.

Functional grasp is essential for enabling dexterous multi-finger robot hands to manipulate objects effectively. However, most prior work either focuses on power grasping, which simply involves holding an object still, or relies on costly teleoperated robot demonstrations to teach robots how to grasp each object functionally. Instead, we propose extracting human grasp information from web images since they depict natural and functional object interactions, thereby bypassing the need for curated demonstrations. We reconstruct human hand-object interaction (HOI) 3D meshes from RGB images, retarget the human hand to multi-finger robot hands, and align the noisy object mesh with its accurate 3D shape. We show that these relatively low-quality HOI data from inexpensive web sources can effectively train a functional grasping model. To further expand the grasp dataset for seen and unseen objects, we use the initially-trained grasping policy with web data in the IsaacGym simulator to generate physically feasible grasps while preserving functionality. We train the grasping model on 10 object categories and evaluate it on 9 unseen objects, including challenging items such as syringes, pens, spray bottles, and tongs, which are underrepresented in existing datasets. The model trained on the web HOI dataset, achieving a 75.8% success rate on seen objects and 61.8% across all objects in simulation, with a 6.7% improvement in success rate and a 1.8x increase in functionality ratings over baselines. Simulator-augmented data further boosts performance from 61.8% to 83.4%. The sim-to-real transfer to the LEAP Hand achieves a 85% success rate. Project website is at: https://web2grasp.github.io/.

Yufei Wang, Ziyu Wang, Mino Nakura et al. · cmu

This paper presents ArticuBot, in which a single learned policy enables a robotics system to open diverse categories of unseen articulated objects in the real world. This task has long been challenging for robotics due to the large variations in the geometry, size, and articulation types of such objects. Our system, Articubot, consists of three parts: generating a large number of demonstrations in physics-based simulation, distilling all generated demonstrations into a point cloud-based neural policy via imitation learning, and performing zero-shot sim2real transfer to real robotics systems. Utilizing sampling-based grasping and motion planning, our demonstration generalization pipeline is fast and effective, generating a total of 42.3k demonstrations over 322 training articulated objects. For policy learning, we propose a novel hierarchical policy representation, in which the high-level policy learns the sub-goal for the end-effector, and the low-level policy learns how to move the end-effector conditioned on the predicted goal. We demonstrate that this hierarchical approach achieves much better object-level generalization compared to the non-hierarchical version. We further propose a novel weighted displacement model for the high-level policy that grounds the prediction into the existing 3D structure of the scene, outperforming alternative policy representations. We show that our learned policy can zero-shot transfer to three different real robot settings: a fixed table-top Franka arm across two different labs, and an X-Arm on a mobile base, opening multiple unseen articulated objects across two labs, real lounges, and kitchens. Videos and code can be found on our project website: https://articubot.github.io/.

Alexis Yihong Hao, Yufei Wang, Navin Sriram Ravie et al. · cmu

Robot-assisted dressing has the potential to significantly improve the lives of individuals with mobility impairments. To ensure an effective and comfortable dressing experience, the robot must be able to handle challenging deformable garments, apply appropriate forces, and adapt to limb movements throughout the dressing process. Prior work often makes simplifying assumptions -- such as static human limbs during dressing -- which limits real-world applicability. In this work, we develop a robot-assisted dressing system capable of handling partial observations with visual occlusions, as well as robustly adapting to arm motions during the dressing process. Given a policy trained in simulation with partial observations, we propose a method to fine-tune it in the real world using a small amount of data and multi-modal feedback from vision and force sensing, to further improve the policy's adaptability to arm motions and enhance safety. We evaluate our method in simulation with simplified articulated human meshes and in a real world human study with 12 participants across 264 dressing trials. Our policy successfully dresses two long-sleeve everyday garments onto the participants while being adaptive to various kinds of arm motions, and greatly outperforms prior baselines in terms of task completion and user feedback. Video are available at https://dressing-motion.github.io/.

Divyam Goel, Yufei Wang, Tiancheng Wu et al. · cmu

Standard evaluation protocols in robotic manipulation typically assess policy performance over curated, in-distribution test sets, offering limited insight into how systems fail under plausible variation. We introduce Geometric Red-Teaming (GRT), a red-teaming framework that probes robustness through object-centric geometric perturbations, automatically generating CrashShapes -- structurally valid, user-constrained mesh deformations that trigger catastrophic failures in pre-trained manipulation policies. The method integrates a Jacobian field-based deformation model with a gradient-free, simulator-in-the-loop optimization strategy. Across insertion, articulation, and grasping tasks, GRT consistently discovers deformations that collapse policy performance, revealing brittle failure modes missed by static benchmarks. By combining task-level policy rollouts with constraint-aware shape exploration, we aim to build a general purpose framework for structured, object-centric robustness evaluation in robotic manipulation. We additionally show that fine-tuning on individual CrashShapes, a process we refer to as blue-teaming, improves task success by up to 60 percentage points on those shapes, while preserving performance on the original object, demonstrating the utility of red-teamed geometries for targeted policy refinement. Finally, we validate both red-teaming and blue-teaming results with a real robotic arm, observing that simulated CrashShapes reduce task success from 90% to as low as 22.5%, and that blue-teaming recovers performance to up to 90% on the corresponding real-world geometry -- closely matching simulation outcomes. Videos and code can be found on our project website: https://georedteam.github.io/ .

Akhil Padmanabha, Arpit Agarwal, Catherine Li et al.

Three-dimensional (3-D) skin surface reconstruction offers promise for objective and quantitative dermatological assessment, but no portable, high-resolution device exists that has been validated and used for depth reconstruction across various body locations. We present a compact 3-D skin reconstruction probe based on GelSight tactile imaging with a custom elastic gel and a learning-based reconstruction algorithm for micron-level wrinkle height estimation. Our probe, integrated into a handheld probe with force sensing for consistent contact, achieves a mean absolute error of 12.55 micron on wrinkle-like test objects. In a study with 15 participants without skin disorders, we provide the first validated wrinkle depth metrics across multiple body regions. We further demonstrate statistically significant reductions in wrinkle height at three locations following over-the-counter moisturizer application. Our work offers a validated tool for clinical and cosmetic skin analysis, with potential applications in diagnosis, treatment monitoring, and skincare efficacy evaluation.

Jing Gao, Ce Zheng, Laszlo A. Jeni et al.

In-bed human mesh recovery can be crucial and enabling for several healthcare applications, including sleep pattern monitoring, rehabilitation support, and pressure ulcer prevention. However, it is difficult to collect large real-world visual datasets in this domain, in part due to privacy and expense constraints, which in turn presents significant challenges for training and deploying deep learning models. Existing in-bed human mesh estimation methods often rely heavily on real-world data, limiting their ability to generalize across different in-bed scenarios, such as varying coverings and environmental settings. To address this, we propose a Sim-to-Real Transfer Framework for in-bed human mesh recovery from overhead depth images, which leverages large-scale synthetic data alongside limited or no real-world samples. We introduce a diffusion model that bridges the gap between synthetic data and real data to support generalization in real-world in-bed pose and body inference scenarios. Extensive experiments and ablation studies validate the effectiveness of our framework, demonstrating significant improvements in robustness and adaptability across diverse healthcare scenarios.

Fukang Liu, Vaidehi Patil, Zackory Erickson et al.

Robotic bathing assistance has long been considered an important and practical task in healthcare. Yet, achieving flexible and efficient cleaning tasks on the human body is challenging, since washing the body involves direct human-robot physical contact and simple, safe, and effective devices are needed for bathing and hygiene. In this paper, we present a meso-scale wearable robot that can locomote along the human body to provide bathing and skin care assistance. We evaluated the cleaning performance of the robot system under different scenarios. The experiments on the pipe show that the robot can achieve cleaning percentage over 92% with two types of stretchable fabrics. The robot removed most of the debris with average values of 94% on a human arm and 93% on a manikin torso. The results demonstrate that the robot exhibits high performance in cleaning tasks.

Zackory Erickson, Henry M. Clever, Vamsee Gangaram et al.

Towards the goal of robots performing robust and intelligent physical interactions with people, it is crucial that robots are able to accurately sense the human body, follow trajectories around the body, and track human motion. This study introduces a capacitive servoing control scheme that allows a robot to sense and navigate around human limbs during close physical interactions. Capacitive servoing leverages temporal measurements from a multi-electrode capacitive sensor array mounted on a robot's end effector to estimate the relative position and orientation (pose) of a nearby human limb. Capacitive servoing then uses these human pose estimates from a data-driven pose estimator within a feedback control loop in order to maneuver the robot's end effector around the surface of a human limb. We provide a design overview of capacitive sensors for human-robot interaction and then investigate the performance and generalization of capacitive servoing through an experiment with 12 human participants. The results indicate that multidimensional capacitive servoing enables a robot's end effector to move proximally or distally along human limbs while adapting to human pose. Using a cross-validation experiment, results further show that capacitive servoing generalizes well across people with different body size.

Zackory Erickson, Yijun Gu, Charles C. Kemp

Versatile robotic caregivers could benefit millions of people worldwide, including older adults and people with disabilities. Recent work has explored how robotic caregivers can learn to interact with people through physics simulations, yet transferring what has been learned to real robots remains challenging. Virtual reality (VR) has the potential to help bridge the gap between simulations and the real world. We present Assistive VR Gym (AVR Gym), which enables real people to interact with virtual assistive robots. We also provide evidence that AVR Gym can help researchers improve the performance of simulation-trained assistive robots with real people. Prior to AVR Gym, we trained robot control policies (Original Policies) solely in simulation for four robotic caregiving tasks (robot-assisted feeding, drinking, itch scratching, and bed bathing) with two simulated robots (PR2 from Willow Garage and Jaco from Kinova). With AVR Gym, we developed Revised Policies based on insights gained from testing the Original policies with real people. Through a formal study with eight participants in AVR Gym, we found that the Original policies performed poorly, the Revised policies performed significantly better, and that improvements to the biomechanical models used to train the Revised policies resulted in simulated people that better match real participants. Notably, participants significantly disagreed that the Original policies were successful at assistance, but significantly agreed that the Revised policies were successful at assistance. Overall, our results suggest that VR can be used to improve the performance of simulation-trained control policies with real people without putting people at risk, thereby serving as a valuable stepping stone to real robotic assistance.

Henry M. Clever, Zackory Erickson, Ariel Kapusta et al.

People spend a substantial part of their lives at rest in bed. 3D human pose and shape estimation for this activity would have numerous beneficial applications, yet line-of-sight perception is complicated by occlusion from bedding. Pressure sensing mats are a promising alternative, but training data is challenging to collect at scale. We describe a physics-based method that simulates human bodies at rest in a bed with a pressure sensing mat, and present PressurePose, a synthetic dataset with 206K pressure images with 3D human poses and shapes. We also present PressureNet, a deep learning model that estimates human pose and shape given a pressure image and gender. PressureNet incorporates a pressure map reconstruction (PMR) network that models pressure image generation to promote consistency between estimated 3D body models and pressure image input. In our evaluations, PressureNet performed well with real data from participants in diverse poses, even though it had only been trained with synthetic data. When we ablated the PMR network, performance dropped substantially.

Zackory Erickson, Eliot Xing, Bharat Srirangam et al.

Material recognition can help inform robots about how to properly interact with and manipulate real-world objects. In this paper, we present a multimodal sensing technique, leveraging near-infrared spectroscopy and close-range high resolution texture imaging, that enables robots to estimate the materials of household objects. We release a dataset of high resolution texture images and spectral measurements collected from a mobile manipulator that interacted with 144 household objects. We then present a neural network architecture that learns a compact multimodal representation of spectral measurements and texture images. When generalizing material classification to new objects, we show that this multimodal representation enables a robot to recognize materials with greater performance as compared to prior state-of-the-art approaches. Finally, we present how a robot can combine this high resolution local sensing with images from the robot's head-mounted camera to achieve accurate material classification over a scene of objects on a table.

Alexander Clegg, Zackory Erickson, Patrick Grady et al.

We investigated the application of haptic feedback control and deep reinforcement learning (DRL) to robot-assisted dressing. Our method uses DRL to simultaneously train human and robot control policies as separate neural networks using physics simulations. In addition, we modeled variations in human impairments relevant to dressing, including unilateral muscle weakness, involuntary arm motion, and limited range of motion. Our approach resulted in control policies that successfully collaborate in a variety of simulated dressing tasks involving a hospital gown and a T-shirt. In addition, our approach resulted in policies trained in simulation that enabled a real PR2 robot to dress the arm of a humanoid robot with a hospital gown. We found that training policies for specific impairments dramatically improved performance; that controller execution speed could be scaled after training to reduce the robot's speed without steep reductions in performance; that curriculum learning could be used to lower applied forces; and that multi-modal sensing, including a simulated capacitive sensor, improved performance.

Daehyung Park, Yuuna Hoshi, Harshal P. Mahajan et al.

Eating is an essential activity of daily living (ADL) for staying healthy and living at home independently. Although numerous assistive devices have been introduced, many people with disabilities are still restricted from independent eating due to the devices' physical or perceptual limitations. In this work, we present a new meal-assistance system and evaluations of this system with people with motor impairments. We also discuss learned lessons and design insights based on the evaluations. The meal-assistance system uses a general-purpose mobile manipulator, a Willow Garage PR2, which has the potential to serve as a versatile form of assistive technology. Our active feeding framework enables the robot to autonomously deliver food to the user's mouth, reducing the need for head movement by the user. The user interface, visually-guided behaviors, and safety tools allow people with severe motor impairments to successfully use the system. We evaluated our system with a total of 10 able-bodied participants and 9 participants with motor impairments. Both groups of participants successfully ate various foods using the system and reported high rates of success for the system's autonomous behaviors. In general, participants who operated the system reported that it was comfortable, safe, and easy-to-use.

Zackory Erickson, Henry M. Clever, Vamsee Gangaram et al.

Robotic assistance presents an opportunity to benefit the lives of many people with physical disabilities, yet accurately sensing the human body and tracking human motion remain difficult for robots. We present a multidimensional capacitive sensing technique that estimates the local pose of a human limb in real time. A key benefit of this sensing method is that it can sense the limb through opaque materials, including fabrics and wet cloth. Our method uses a multielectrode capacitive sensor mounted to a robot's end effector. A neural network model estimates the position of the closest point on a person's limb and the orientation of the limb's central axis relative to the sensor's frame of reference. These pose estimates enable the robot to move its end effector with respect to the limb using feedback control. We demonstrate that a PR2 robot can use this approach with a custom six electrode capacitive sensor to assist with two activities of daily living-dressing and bathing. The robot pulled the sleeve of a hospital gown onto able-bodied participants' right arms, while tracking human motion. When assisting with bathing, the robot moved a soft wet washcloth to follow the contours of able-bodied participants' limbs, cleaning their surfaces. Overall, we found that multidimensional capacitive sensing presents a promising approach for robots to sense and track the human body during assistive tasks that require physical human-robot interaction.

Zackory Erickson, Nathan Luskey, Sonia Chernova et al.

Recognizing an object's material can inform a robot on the object's fragility or appropriate use. To estimate an object's material during manipulation, many prior works have explored the use of haptic sensing. In this paper, we explore a technique for robots to estimate the materials of objects using spectroscopy. We demonstrate that spectrometers provide several benefits for material recognition, including fast response times and accurate measurements with low noise. Furthermore, spectrometers do not require direct contact with an object. To explore this, we collected a dataset of spectral measurements from two commercially available spectrometers during which a robotic platform interacted with 50 flat material objects, and we show that a neural network model can accurately analyze these measurements. Due to the similarity between consecutive spectral measurements, our model achieved a material classification accuracy of 94.6% when given only one spectral sample per object. Similar to prior works with haptic sensors, we found that generalizing material recognition to new objects posed a greater challenge, for which we achieved an accuracy of 79.1% via leave-one-object-out cross-validation. Finally, we demonstrate how a PR2 robot can leverage spectrometers to estimate the materials of everyday objects found in the home. From this work, we find that spectroscopy poses a promising approach for material classification during robotic manipulation.

Henry M. Clever, Ariel Kapusta, Daehyung Park et al.

Robots have the potential to assist people in bed, such as in healthcare settings, yet bedding materials like sheets and blankets can make observation of the human body difficult for robots. A pressure-sensing mat on a bed can provide pressure images that are relatively insensitive to bedding materials. However, prior work on estimating human pose from pressure images has been restricted to 2D pose estimates and flat beds. In this work, we present two convolutional neural networks to estimate the 3D joint positions of a person in a configurable bed from a single pressure image. The first network directly outputs 3D joint positions, while the second outputs a kinematic model that includes estimated joint angles and limb lengths. We evaluated our networks on data from 17 human participants with two bed configurations: supine and seated. Our networks achieved a mean joint position error of 77 mm when tested with data from people outside the training set, outperforming several baselines. We also present a simple mechanical model that provides insight into ambiguity associated with limbs raised off of the pressure mat, and demonstrate that Monte Carlo dropout can be used to estimate pose confidence in these situations. Finally, we provide a demonstration in which a mobile manipulator uses our network's estimated kinematic model to reach a location on a person's body in spite of the person being seated in a bed and covered by a blanket.

Zackory Erickson, Henry M. Clever, Greg Turk et al.

Robot-assisted dressing offers an opportunity to benefit the lives of many people with disabilities, such as some older adults. However, robots currently lack common sense about the physical implications of their actions on people. The physical implications of dressing are complicated by non-rigid garments, which can result in a robot indirectly applying high forces to a person's body. We present a deep recurrent model that, when given a proposed action by the robot, predicts the forces a garment will apply to a person's body. We also show that a robot can provide better dressing assistance by using this model with model predictive control. The predictions made by our model only use haptic and kinematic observations from the robot's end effector, which are readily attainable. Collecting training data from real world physical human-robot interaction can be time consuming, costly, and put people at risk. Instead, we train our predictive model using data collected in an entirely self-supervised fashion from a physics-based simulation. We evaluated our approach with a PR2 robot that attempted to pull a hospital gown onto the arms of 10 human participants. With a 0.2s prediction horizon, our controller succeeded at high rates and lowered applied force while navigating the garment around a persons fist and elbow without getting caught. Shorter prediction horizons resulted in significantly reduced performance with the sleeve catching on the participants' fists and elbows, demonstrating the value of our model's predictions. These behaviors of mitigating catches emerged from our deep predictive model and the controller objective function, which primarily penalizes high forces.

Zackory Erickson, Maggie Collier, Ariel Kapusta et al.

Dressing is a fundamental task of everyday living and robots offer an opportunity to assist people with motor impairments. While several robotic systems have explored robot-assisted dressing, few have considered how a robot can manage errors in human pose estimation, or adapt to human motion in real time during dressing assistance. In addition, estimating pose changes due to human motion can be challenging with vision-based techniques since dressing is often intended to visually occlude the body with clothing. We present a method to track a person's pose in real time using capacitive proximity sensing. This sensing approach gives direct estimates of distance with low latency, has a high signal-to-noise ratio, and has low computational requirements. Using our method, a robot can adjust for errors in the estimated pose of a person and physically follow the contours and movements of the person while providing dressing assistance. As part of an evaluation of our method, the robot successfully pulled the sleeve of a hospital gown and a cardigan onto the right arms of 10 human participants, despite arm motions and large errors in the initially estimated pose of the person's arm. We also show that a capacitive sensor is unaffected by visual occlusion of the body and can sense a person's body through cotton clothing.

Zackory Erickson, Sonia Chernova, Charles C. Kemp

Material recognition enables robots to incorporate knowledge of material properties into their interactions with everyday objects. For example, material recognition opens up opportunities for clearer communication with a robot, such as "bring me the metal coffee mug", and recognizing plastic versus metal is crucial when using a microwave or oven. However, collecting labeled training data with a robot is often more difficult than unlabeled data. We present a semi-supervised learning approach for material recognition that uses generative adversarial networks (GANs) with haptic features such as force, temperature, and vibration. Our approach achieves state-of-the-art results and enables a robot to estimate the material class of household objects with ~90% accuracy when 92% of the training data are unlabeled. We explore how well this approach can recognize the material of new objects and we discuss challenges facing generalization. To motivate learning from unlabeled training data, we also compare results against several common supervised learning classifiers. In addition, we have released the dataset used for this work which consists of time-series haptic measurements from a robot that conducted thousands of interactions with 72 household objects.

Alexander Clegg, Wenhao Yu, Zackory Erickson et al.

We present a controller that allows an arm-like manipulator to navigate deformable cloth garments in simulation through the use of haptic information. The main challenge of such a controller is to avoid getting tangled in, tearing or punching through the deforming cloth. Our controller aggregates force information from a number of haptic-sensing spheres all along the manipulator for guidance. Based on haptic forces, each individual sphere updates its target location, and the conflicts that arise between this set of desired positions is resolved by solving an inverse kinematic problem with constraints. Reinforcement learning is used to train the controller for a single haptic-sensing sphere, where a training run is terminated (and thus penalized) when large forces are detected due to contact between the sphere and a simplified model of the cloth. In simulation, we demonstrate successful navigation of a robotic arm through a variety of garments, including an isolated sleeve, a jacket, a shirt, and shorts. Our controller out-performs two baseline controllers: one without haptics and another that was trained based on large forces between the sphere and cloth, but without early termination.