Meta tags:
description=;
author=;

Headings (most frequently used words):

to, accepted, is, 2024, 2025, 2022, 08, and, 2023, on, with, manipulation, corl, rlvlmf, iros, 04, rss, 09, review, paper, visual, policy, ra, at, news, publications, three, papers, robot, assisted, dressing, arm, motions, geometric, red, teaming, for, robotic, policies, articulated, asset, generation, have, been, 06, offline, articubot, difftori, neurips, as, spotlight, our, deformable, object, unfolding, the, literature, annual, of, control, robotics, autonomous, systems, 05, robogen, are, icml, 02, force, constrained, did, wonderful, summer, intern, mit, ibm, watson, ai, lab, hosted, by, dr, chuang, gan, one, dress, them, all, toolflownet, 07, haptic, reasoning, presentation,

Text of the page (most frequently used words):
and (284), the (200), learning (94), with (70), for (62), that (60), wang (59), policy (53), yufei (45), robot (39), our (38), david (38), held (36), paper (33), #manipulation (33), reinforcement (32), from (31), real (30), this (29), model (28), dressing (28), based (27), zackory (27), erickson (27), can (24), object (24), abstract (23), robotic (23), which (22), tasks (22), cloth (22), year (21), title (20), author (20), bibtex (20), simulation (20), visual (19), conference (18), task (18), objects (18), world (18), deformable (17), robotics (17), project (17), arm (17), 2024 (17), articulated (17), 2025 (17), inproceedings (16), deep (16), such (16), method (16), reward (16), dynamics (16), booktitle (15), state (15), page (15), propose (14), methods (14), learn (14), human (14), code (14), via (14), corl (14), models (14), assisted (14), force (14), vision (13), show (13), these (13), has (12), function (12), optimization (12), data (12), point (12), system (12), work (11), are (11), 2020 (11), policies (11), reasoning (11), learned (10), approach (10), trained (10), different (10), across (10), level (10), poses (10), prior (10), systems (10), accepted (10), information (9), new (9), novel (9), use (9), language (9), how (9), return (9), optimal (9), challenging (9), imitation (9), observations (9), goal (9), flow (9), both (9), then (9), 2022 (9), using (9), textiles (9), garments (9), trajectory (9), offline (9), been (8), long (8), process (8), well (8), research (8), automatically (8), sac (8), demonstrate (8), high (8), representation (8), performance (8), effective (8), 2023 (8), feedback (8), teaming (8), proposed (7), framework (7), training (7), machine (7), one (7), outperforms (7), over (7), control (7), agent (7), rss (7), open (7), textile (7), dress (7), diverse (7), large (7), generative (7), previous (6), first (6), laws (6), dimensional (6), meta (6), end (6), two (6), constrained (6), expert (6), present (6), used (6), transfer (6), xingyu (6), lin (6), due (6), existing (6), further (6), challenges (6), learns (6), partial (6), ieee (6), elastic (6), modeling (6), sun (6), zhanyi (6), science (6), into (6), xian (6), zhou (6), chen (6), review (6), difftop (6), ziyu (6), red (6), motions (6), security (5), games (5), time (5), proceedings (5), actions (5), algorithm (5), solving (5), through (5), uses (5), several (5), conservation (5), many (5), including (5), physics (5), success (5), works (5), network (5), irl (5), behavior (5), gradient (5), its (5), low (5), image (5), rigid (5), enable (5), context (5), self (5), fabricflownet (5), also (5), introduce (5), significantly (5), cloud (5), haptic (5), forces (5), journal (5), automation (5), people (5), enables (5), physical (5), interaction (5), iros (5), toolflownet (5), tools (5), clouds (5), datasets (5), sensing (5), participants (5), vlm (5), robogen (5), gan (5), chuang (5), scale (5), sub (5), demonstrations (5), articubot (5), green (4), volume (4), number (4), pages (4), however (4), upon (4), numerical (4), will (4), problem (4), current (4), finally (4), compared (4), art (4), supervised (4), future (4), rewards (4), key (4), results (4), functions (4), automated (4), between (4), entropy (4), limitations (4), benchmarking (4), icml (4), where (4), directly (4), corresponding (4), benchmarks (4), approaches (4), complex (4), environments (4), algorithms (4), experiments (4), roll (4), without (4), improve (4), cmu (4), 2021 (4), address (4), technique (4), single (4), free (4), when (4), shapes (4), visible (4), mesh (4), article (4), interactions (4), letters (4), publisher (4), during (4), body (4), while (4), progress (4), neural (4), baselines (4), longhini (4), alberta (4), moletta (4), marco (4), welle (4), michael (4), various (4), them (4), all (4), develop (4), able (4), study (4), safety (4), pmlr (4), foundation (4), international (4), often (4), generates (4), tsun (4), hsuan (4), yian (4), generate (4), field (4), differentiable (4), ravie (4), navin (4), sriram (4), articulate (4), anymesh (4), vocabulary (4), geometric (4), prof (4), 2019 (3), have (3), applied (3), domains (3), game (3), sequential (3), dedol (3), strategies (3), applications (3), engineering (3), computational (3), recent (3), computer (3), solvers (3), manner (3), furthermore (3), designed (3), makes (3), given (3), objective (3), insight (3), could (3), form (3), enhance (3), towards (3), general (3), employ (3), neurips (3), workshop (3), automatic (3), tianwei (3), exploration (3), temperature (3), tune (3), achieves (3), promising (3), most (3), video (3), inverse (3), matching (3), cost (3), density (3), behaviors (3), generalization (3), softgym (3), driven (3), only (3), simulated (3), standard (3), evaluate (3), performing (3), specifically (3), segmentation (3), scene (3), better (3), generation (3), loss (3), available (3), google (3), bimanual (3), motion (3), leverages (3), cloths (3), connectivity (3), remains (3), underlying (3), generalize (3), greatly (3), zero (3), shot (3), successfully (3), geometries (3), presentation (3), contact (3), observing (3), individuals (3), assistive (3), tool (3), geometry (3), significant (3), scooping (3), elasticity (3), kragic (3), danica (3), material (3), properties (3), construction (3), commonly (3), accurate (3), aim (3), assumptions (3), fixed (3), range (3), trials (3), multi (3), modal (3), deformations (3), planning (3), requires (3), text (3), description (3), supervision (3), skills (3), skill (3), unfolding (3), literature (3), perception (3), three (3), parts (3), generating (3), hierarchical (3), robots (3), dataset (3), qiu (3), crashshapes (3), blue (3), advised (3), university (3), zheyuan (2), ryan (2), shi (2), lantao (2), rohit (2), singh (2), lucas (2), joppa (2), fei (2), fang (2), aaai (2), gsgs (2), conducted (2), footprints (2), agents (2), movement (2), design (2), strategy (2), best (2), built (2), oracle (2), space (2), response (2), structure (2), specific (2), local (2), modes (2), knowledge (2), bin (2), dong (2), nature (2), associated (2), mathematics (2), especially (2), processing (2), community (2), combining (2), view (2), concept (2), focus (2), scalar (2), deploy (2), should (2), spatial (2), adaptive (2), schemes (2), way (2), easily (2), good (2), other (2), capable (2), discretize (2), provide (2), performs (2), some (2), generalizes (2), weighted (2), value (2), classic (2), exponentially (2), discounted (2), sum (2), there (2), same (2), derived (2), dramatically (2), test (2), indicate (2), exploitation (2), issue (2), soft (2), actor (2), critic (2), original (2), controls (2), empirically (2), metagradient (2), marginal (2), sikchi (2), harshit (2), gupta (2), tejus (2), lee (2), lisa (2), eysenbach (2), ben (2), result (2), any (2), distribution (2), parameters (2), recovers (2), terms (2), required (2), solve (2), utility (2), hard (2), explore (2), structured (2), improved (2), order (2), dice (2), equal (2), advising (2), jake (2), olkin (2), manipulating (2), challenge (2), provides (2), observability (2), simple (2), those (2), set (2), benchmark (2), important (2), forward (2), videos (2), found (2), website (2), gautham (2), brian (2), okorn (2), typically (2), sampling (2), occlusion (2), scheme (2), final (2), https (2), sites (2), com (2), andrew (2), edu (2), thomas (2), weng (2), sujay (2), bajracharya (2), observation (2), images (2), action (2), demonstrating (2), sim (2), zixuan (2), huang (2), smoothing (2), occlusions (2), particle (2), infer (2), points (2), graph (2), more (2), franka (2), configurations (2), estimating (2), presents (2), washing (2), bathing (2), impairments (2), perform (2), representations (2), evaluations (2), varying (2), sizes (2), mobile (2), predicting (2), seita (2), daniel (2), shetty (2), sarthak (2), edward (2), widely (2), modality (2), despite (2), apply (2), not (2), find (2), encoding (2), reichlin (2), alfredo (2), kravberg (2), alexander (2), icra (2), relies (2), advanced (2), being (2), yarn (2), property (2), integrates (2), non (2), benefit (2), lives (2), adults (2), disabilities (2), potential (2), garment (2), pose (2), their (2), ability (2), proper (2), leverage (2), multiple (2), wide (2), news (2), safe (2), quality (2), ascertain (2), types (2), ensure (2), facilitate (2), v235 (2), zhang (2), jesse (2), erdem (2), 41st (2), editor (2), salakhutdinov (2), ruslan (2), kolter (2), zico (2), heller (2), katherine (2), weller (2), adrian (2), oliver (2), nuria (2), scarlett (2), jonathan (2), berkenkamp (2), felix (2), 235 (2), series (2), month (2), jul (2), extensive (2), effort (2), preference (2), labels (2), outperforming (2), under (2), unleashing (2), infinite (2), feng (2), fragkiadaki (2), katerina (2), advancements (2), adapting (2), assets (2), extract (2), fully (2), pipeline (2), annual (2), autonomous (2), garcia (2), camacho (2), irene (2), blanco (2), mulero (2), guillem (2), yin (2), hang (2), arxiv (2), healthcare (2), industrial (2), interest (2), what (2), restricts (2), variations (2), wan (2), weikang (2), utilizes (2), spotlight (2), difftori (2), universal (2), nakura (2), mino (2), bhowal (2), pratik (2), kuo (2), chia (2), liang (2), ting (2), categories (2), unseen (2), articulation (2), generated (2), grasping (2), effector (2), labs (2), venkataraman (2), sreyas (2), rsj (2), intelligent (2), pre (2), online (2), labeled (2), build (2), applicability (2), xiaowen (2), yang (2), jincheng (2), zhehuan (2), joints (2), limited (2), additionally (2), goel (2), divyam (2), tiancheng (2), qiao (2), helen (2), piliptchak (2), pavel (2), evaluation (2), robustness (2), centric (2), user (2), static (2), fine (2), modulated (2), hao (2), yihong (2), hegde (2), bharath (2), rlvlmf (2), received (2), carnegie (2), mellon (2), fellowship (2), 2026 (2), wang2019deep, artificial, intelligence, 1401, 1408, optimize, patrols, law, enforcement, agencies, combating, poaching, illegal, logging, overfishing, subsequent, receiving, rangers, following, chase, poacher, neglected, fill, gap, gsg, augments, vital, element, second, compute, patrolling, adapts, against, responding, attacker, double, restricted, iteratively, adding, networks, exploring, domain, heuristic, initial, constructs, efficient, parallelized, attempt, slides, ziju, shen, zichao, considered, fundamental, broad, fields, chemistry, biology, geology, differential, equations, major, branch, areas, natural, attracted, lot, attention, inspired, intriguing, traditional, pde, mdp, proof, machinery, train, decide, solutions, approximated, temporal, naturally, viewed, decision, making, enforce, term, accuracy, carefully, determine, discrete, approximation, solution, essentially, words, situation, mimicking, experts, details, weno, l3d, pinn, communications, qiwei, tie, yan, liu, accumulated, expected, serve, guides, defined, feasible, ways, define, thus, yet, forms, might, render, speeds, modify, mathematical, specially, maze, environment, atari, experimental, clearly, advantages, beyond, dilemma, crucial, balance, balances, hence, trade, off, shown, very, sensitive, hyperparameter, follow, adjustment, core, namely, along, performances, mujoco, humanoid, link, auto, firl2020corl, suited, difficult, program, specify, match, main, analytic, divergence, stationary, descent, hand, target, implicitly, adversarial, sample, efficiency, trajectories, moreover, recovered, quickly, downstream, changes, projec, abridged, corl2020softgym, direction, manipulate, cover, direct, relatively, source, openai, gym, api, python, interface, creating, reproducible, area, variety, highlight, dealing, intrinsic, dimensionality, partially, observable, analysis, strengths, help, development, narayan, corl2020roll, narasimhan, operate, whole, leads, inefficient, ineffective, incorporating, unknown, ignore, distractors, computation, employing, auxiliary, faster, weng2021fabricflownet, agrawal, khush, directed, deformability, optical, normally, estimation, ffn, input, elegantly, switches, dual, desired, square, shirts, rectangular, lin2021vcd, lack, contrast, pixel, compressed, latent, vector, overcome, connected, strong, inductive, bias, invariant, features, predictions, visualized, smooth, materials, colors, crumpled, supplement, anonymous, vcd, wang2022visual, 11426, 11433, highly, opportunity, assist, daily, dishes, folding, laundry, hygiene, assistance, severe, motor, formulation, collaborative, act, inferring, location, magnitude, distinct, kinematic, quantitative, dish, manipulator, estimate, physically, seita2022toolflownet, canonical, conveys, classification, predicts, dense, per, derive, transformation, execute, continuous, pouring, attain, yao, longhini2023elastic, internation, dextreous, capabilities, complexity, pulled, stretched, today, adopted, annotated, identification, assessed, affects, intertwined, known, priori, almost, impossible, identify, platforms, affect, definition, stress, strain, curves, reformulated, gnn, generalized, effect, dimension, linear, highlighting, setups, sense, wang2023one, older, involves, but, they, make, certain, limit, architecture, distillation, combine, ranges, conduct, comprehensive, 510, dressed, length, arms, average, sun2024force, profoundly, life, achieve, former, latter, helps, maintain, comfort, modalities, trains, simulating, interact, combines, applying, excessive, person, 240, showing, wang24bn, biyik, 51484, 51501, leveraging, vlms, query, give, preferences, pairs, produces, need, pretrained, bıyık, wang24cc, 51936, 51983, latest, instead, produce, advocate, diversified, scenes, supervisions, thereby, scaling, minimal, equips, guided, cycle, proposes, interesting, populating, pertinent, afterwards, decomposes, selects, acquire, attempts, versatile, embedded, queried, repeatedly, producing, endless, stream, longhini2024unfolding, aleny, others, preprint, 2407, 01361, realm, spans, clothing, households, sports, unique, cannot, increasing, within, led, resulting, tailored, subcategory, understand, hinders, generalizing, broader, overview, focusing, extent, addressed, conclude, identifying, problems, outlining, grand, drive, alenyà, júlia, borràs, wan2024difftop, advances, introduces, powerful, parameterized, computing, gradients, respect, addresses, mismatch, maximize, differentiating, suites, sensory, compare, feed, classes, energy, ebm, diffusion, 35imitation, inputs, wang2025articubot, size, consists, distilling, sim2real, utilizing, demonstration, fast, total, 322, move, conditioned, predicted, much, version, displacement, grounds, prediction, alternative, settings, table, top, base, opening, lounges, kitchens, venkataraman2025real, collected, ideal, critical, scenarios, collecting, slow, costly, risky, assume, already, ground, truth, states, implicit, involving, outperform, cloning, summary, labeling, unlabeled, qiu2025articulate, since, capture, surface, semantically, meaningful, spatially, precise, structures, heavily, depend, handcrafted, cabinets, drawers, convert, counterpart, prompting, techniques, semantic, allowing, functional, toys, mechanical, devices, vehicles, expanding, coverage, acquisition, transferred, goel2025geometric, protocols, assess, curated, sets, offering, fail, plausible, variation, probes, perturbations, structurally, valid, trigger, catastrophic, failures, jacobian, deformation, simulator, loop, insertion, consistently, discovers, collapse, revealing, brittle, failure, missed, rollouts, constraint, aware, shape, purpose, tuning, individual, refer, improves, percentage, preserving, teamed, targeted, refinement, validate, reduce, closely, outcomes, oral, hao2025force, mobility, comfortable, experience, must, handle, appropriate, adapt, limb, movements, throughout, simplifying, limbs, limits, handling, robustly, small, amount, adaptability, simplified, meshes, 264, dresses, sleeve, everyday, onto, kinds, completion, alexis, indicates, contribution, publications, did, wonderful, summer, intern, mit, ibm, watson, lab, hosted, papers, asset, phd, student, fortunate, dec, before, coming, yuanpei, college, july, application, graduate, supported, uber, presidential, softbank, you, background, fit, please, feel, reach, out, yufeiw2, twitter, github, scholar, currently, job, market, academic, positions, starting, fall, peking, department, institute, photo, credit, liying, home, menu,

Text of the page (random words):
odels 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 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 inproceedings pmlr v235 wang24bn title rl vlm f reinforcement learning from vision language foundation model feedback author wang yufei and sun zhanyi and zhang jesse and xian zhou and biyik erdem and held david and erickson zackory booktitle proceedings of the 41st international conference on machine learning pages 51484 51501 year 2024 editor salakhutdinov ruslan and kolter zico and heller katherine and weller adrian and oliver nuria and scarlett jonathan and berkenkamp felix volume 235 series proceedings of machine learning research month 21 27 jul publisher pmlr force constrained visual policy safe robot assisted dressing via multi modal sensing zhanyi sun yufei wang david held zackory erickson robotics and automation letters ra l 2024 paper project page abstract bibtex robot assisted dressing could profoundly enhance the quality of life of adults with physical disabilities to achieve this a robot can benefit from both visual and force sensing the former enables the robot to ascertain human body pose and garment deformations while the latter helps maintain safety and comfort during the dressing process in this paper we introduce a new technique that leverages both vision and force modalities for this assistive task our approach first trains a vision based dressing policy using reinforcement learning in simulation with varying body sizes poses and types of garments we then learn a force dynamics model for action planning to ensure safety due to limitations of simulating accurate force data when deformable garments interact with the human body we learn a force dynamics model directly from real world data our proposed method combines the vision based policy trained in simulation with the force dynamics model learned in the real world by solving a constrained optimization problem to infer actions that facilitate the dressing process without applying excessive force on the person we evaluate our system in simulation and in a real world human study with 10 participants across 240 dressing trials showing it greatly outperforms prior baselines article sun2024force title force constrained visual policy safe robot assisted dressing via multi modal sensing author sun zhanyi and wang yufei and held david and erickson zackory journal ieee robotics and automation letters year 2024 publisher ieee one policy to dress them all learning to dress people with diverse poses and garments yufei wang zhanyi sun zackory erickson david held rss 2023 paper project page cmu news abstract bibtex robot assisted dressing could benefit the lives of many people such as older adults and individuals with disabilities despite such potential robot assisted dressing remains a challenging task for robotics as it involves complex manipulation of deformable cloth in 3d space many prior works aim to solve the robot assisted dressing task but they make certain assumptions such as a fixed garment and a fixed arm pose that limit their ability to generalize in this work we develop a robot assisted dressing system that is able to dress different garments on people with diverse poses from partial point cloud observations based on a learned policy we show that with proper design of the policy architecture and q function reinforcement learning rl can be used to learn effective policies with partial point cloud observations that work well for dressing diverse garments we further leverage policy distillation to combine multiple policies trained on different ranges of human arm poses into a single policy that works over a wide range of different arm poses we conduct comprehensive real world evaluations of our system with 510 dressing trials in a human study with 17 participants with different arm poses and dressed garments our system is able to dress 86 of the length of the participants arms on average inproceedings wang2023one title one policy to dress them all learning to dress people with diverse poses and garments author wang yufei and sun zhanyi and erickson zackory and held david booktitle robotics science and systems rss year 2023 elastic context encoding elasticity for data driven models of textiles alberta longhini marco moletta alfredo reichlin michael c welle alexander kravberg yufei wang david held zackory erickson danica kragic icra 2023 paper abstract bibtex physical interaction with textiles such as assistive dressing relies on advanced dextreous capabilities the underlying complexity in textile behavior when being pulled and stretched is due to both the yarn material properties and the textile construction technique today there are no commonly adopted and annotated datasets on which the various interaction or property identification methods are assessed one important property that affects the interaction is material elasticity that results from both the yarn material and construction technique these two are intertwined and if not known a priori almost impossible to identify through sensing commonly available on robotic platforms we introduce elastic context ec a concept that integrates various properties that affect elastic behavior to enable a more effective physical interaction with textiles the definition of ec relies on stress strain curves commonly used in textile engineering which we reformulated for robotic applications we employ ec using graph neural network gnn to learn generalized elastic behaviors of textiles furthermore we explore the effect the dimension of the ec has on accurate force modeling of non linear real world elastic behaviors highlighting the challenges of current robotic setups to sense textile properties inproceedings longhini2023elastic title elastic context encoding elasticity for data driven models of textiles author longhini alberta and moletta marco and reichlin alfredo and welle michael c and kravberg alexander and wang yufei and held david and erickson zackory and kragic danica journal internation conference on robot automation icra year 2023 toolflownet robotic manipulation with tools via predicting tool flow from point clouds daniel seita yufei wang sarthak j shetty edward yao li zackory erickson david held corl 2022 paper project page code abstract bibtex point clouds are a widely available and canonical data modality which conveys the 3d geometry of a scene despite significant progress in classification and segmentation from point clouds policy learning from such a modality remains challenging and most prior works in imitation learning focus on learning policies from images or state information in this paper we propose a novel framework for learning policies from point clouds for robotic manipulation with tools we use a novel neural network toolflownet which predicts dense per point flow on the tool that the robot controls and then uses the flow to derive the transformation that the robot should execute we apply this framework to imitation learning of challenging deformable object manipulation tasks with continuous movement of tools including scooping and pouring and demonstrate significantly improved performance over baselines which do not use flow we perform physical scooping experiments with toolflownet and find that we can attain 78 scooping success inproceedings seita2022toolflownet title toolflownet robotic manipulation with tools via predicting tool flow from point clouds author seita daniel and wang yufei and shetty sarthak and li edward and erickson zackory and held david booktitle conference on robot learning year 2022 visual haptic reasoning estimating contact forces by observing deformable object interactions yufei wang david held zackory erickson robotics and automation letters ra l with presentation at iros 2022 paper project page abstract bibtex 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 article wang2022visual title visual haptic reasoning estimating contact forces by observing deformable object interactions author wang yufei and held david and erickson zackory journal ieee robotics and automation letters volume 7 number 4 pages 11426 11433 year 2022 publisher ieee learning visible connectivity dynamics for cloth smoothing xingyu lin yufei wang zixuan huang david held order by dice corl 2021 paper project page code abstract bibtex robotic manipulation of cloth remains challenging due to the complex dynamics of cloth lack of a low dimensional state representation and self occlusions in contrast to previous model based approaches that learn a pixel based dynamics model or a compressed latent vector dynamics we propose to learn a particle based dynamics model from a partial point cloud observation to overcome the challenges of partial observability we infer which visible points are connected on the underlying cloth mesh we then learn a dynamics model over this visible connectivity graph compared to previous learning based approaches our model poses strong inductive bias with its particle based representation for learning the underlying cloth physics it can generalize to cloths with novel shapes it is invariant to visual features and the predictions can be more easily visualized we show that our method greatly outperforms previous state of the art model based and model free reinforcement learning methods in simulation furthermore we demonstrate zero shot sim to real transfer where we deploy the model trained in simulation on a franka arm and show that the model can successfully smooth cloths of different materials geometries and colors from crumpled configurations videos can be found in the supplement and on our anonymous project website https sites google com view vcd cloth inproceedings lin2021vcd title learning visible connectivity dynamics for cloth smoothing author lin xingyu and wang yufei and huang zixuan and held david booktitle conference on robot learning year 2021 fabricflownet bimanual cloth manipulation with a flow based policy thomas weng sujay bajracharya yufei wang david held corl 2021 paper project page code abstract bibtex we address the problem of goal directed cloth manipulation a challenging task due to the deformability of cloth our insight is that optical flow a technique normally used for motion estimation in video can also provide an effective representation for corresponding cloth poses across observation and goal images we introduce fabricflownet ffn a cloth manipulation policy that leverages flow as both an input and as an action representation to improve performance fabricflownet also elegantly switches between dual arm and single arm actions based on the desired goal we show that fabricflownet significantly outperforms state of the art model free and model based cloth manipulation policies we also present real world experiments on a bimanual system demonstrating effective sim to real transfer finally we show that our method generalizes when trained on a single square cloth to other cloth shapes such as t shirts and rectangular cloths inproceedings weng2021fabricflownet title fabricflownet bimanual cloth manipulation with a flow based policy author weng thomas and bajracharya sujay and wang yufei and agrawal khush and held david booktitle conference on robot learning year 2021 roll visual self supervised reinforcement learning with object reasoning yufei wang gautham narayan xingyu lin brian okorn david held corl 2020 paper project page code abstract bibtex current image based reinforcement learning rl algorithms typically operate on the whole image without performing object level reasoning this leads to inefficient goal sampling and ineffective reward functions in this paper we improve upon previous visual self supervised rl by incorporating object level reasoning and occlusion reasoning specifically we use unknown object segmentation to ignore distractors in the scene for better reward computation and goal generation we further enable occlusion reasoning by employing a novel auxiliary loss and training scheme we demonstrate that our proposed algorithm roll reinforcement learning with object level learning learns dramatically faster and achieves better final performance compared with previous methods in several simulated visual control tasks project video and code are available at https sites google com andrew cmu edu roll inproceedings corl2020roll title roll visual self supervised reinforcement learning with object reasoning author wang yufei and narasimhan gautham and lin xingyu and okorn brian and held david booktitle conference on robot learning year 2020 softgym benchmarking deep reinforcement learning for deformable object manipulation xingyu lin yufei wang jake olkin david held corl 2020 paper project page code abstract bibtex manipulating deformable objects has long been a challenge in robotics due to its high dimensional state representation and complex dynamics recent success in deep reinforcement learning provides a promising direction for learning to manipulate deformable objects with data driven methods however existing reinforcement learning benchmarks only cover tasks with direct state observability and simple low dimensional dynamics or with relatively simple image based environments such as those with rigid objects in this paper we present softgym a set of open source simulated benchmarks for manipulating deformable objects with a standard openai gym api and a python interface for creating new environments our benchmark will enable reproducible research in this im...