The ability to learn from large batches of autonomously collected data for policy improvement -- a paradigm we refer to as batch online reinforcement learning -- holds the promise of enabling truly scalable robot learning by significantly reducing the need for human effort of data collection while getting benefits from self-improvement. Yet, despite the promise of this paradigm, it remains challenging to achieve due to algorithms not being able to learn effectively from the autonomous data. For example, prior works have applied imitation learning and filtered imitation learning methods to the batch online RL problem, but these algorithms often fail to efficiently improve from the autonomously collected data or converge quickly to a suboptimal point. This raises the question of what matters for effective batch online RL in robotics. Motivated by this question, we perform a systematic empirical study of three axes -- (i) algorithm class, (ii) policy extraction methods, and (iii) policy expressivity -- and analyze how these axes affect performance and scaling with the amount of autonomous data. Through our analysis, we make several observations. First, we observe that the use of Q-functions to guide batch online RL significantly improves performance over imitation-based methods. Building on this, we show that an implicit method of policy extraction -- via choosing the best action in the distribution of the policy -- is necessary over traditional policy extraction methods from offline RL. Next, we show that an expressive policy class is preferred over less expressive policy classes. Based on this analysis, we propose a general recipe for effective batch online RL. We then show a simple addition to the recipe of using temporally-correlated noise to obtain more diversity results in further performance gains. Our recipe obtains significantly better performance and scaling compared to prior methods.
Stanford University（斯坦福大学）；Institute for Computational and Mathematical Engineering（计算与数学工程研究所）
Several practical applications of reinforcement learning involve an agent learning from past data without the possibility of further exploration. Often these applications require us to 1) identify a near optimal policy or to 2) estimate the value of a target policy. For both tasks we derive \emph{exponential} information-theoretic lower bounds in discounted infinite horizon MDPs with a linear function representation for the action value function even if 1) \emph{realizability} holds, 2) the batch algorithm observes the exact reward and transition \emph{functions}, and 3) the batch algorithm is…
Learning from diverse offline datasets is a promising path towards learning general purpose robotic agents. However, a core challenge in this paradigm lies in collecting large amounts of meaningful data, while not depending on a human in the loop for data collection. One way to address this challenge is through task-agnostic exploration, where an agent attempts to explore without a task-specific reward function, and collect data that can be useful for any downstream task. While these approaches have shown some promise in simple domains, they often struggle to explore the relevant regions of th…
Princeton University（普林斯顿大学）；University of Pennsylvania（宾夕法尼亚大学）
Reinforcement learning (RL) is a general and well-known method that a robot can use to learn an optimal control policy to solve a particular task. We would like to build a versatile robot that can learn multiple tasks, but using RL for each of them would be prohibitively expensive in terms of both time and wear-and-tear on the robot. To remedy this problem, we use the Policy Gradient Efficient Lifelong Learning Algorithm (PG-ELLA), an online multi-task RL algorithm that enables the robot to efficiently learn multiple consecutive tasks by sharing knowledge between these tasks to accelerate lear…
Courant Institute of Mathematical Sciences, New York University（纽约大学库朗数学科学研究所）；New York University（纽约大学）
Reinforcement learning algorithms have had tremendous successes in online learning settings. However, these successes have relied on low-stakes interactions between the algorithmic agent and its environment. In many settings where RL could be of use, such as health care and autonomous driving, the mistakes made by most online RL algorithms during early training come with unacceptable costs. These settings require developing reinforcement learning algorithms that can operate in the so-called batch setting, where the algorithms must learn from set of data that is fixed, finite, and generated fro…
Recommender System (RS) is an important online application that affects billions of users every day. The mainstream RS ranking framework is composed of two parts: a Multi-Task Learning model (MTL) that predicts various user feedback, i.e., clicks, likes, sharings, and a Multi-Task Fusion model (MTF) that combines the multi-task outputs into one final ranking score with respect to user satisfaction. There has not been much research on the fusion model while it has great impact on the final recommendation as the last crucial process of the ranking. To optimize long-term user satisfaction rather …
A key challenge in Imitation Learning (IL) is that optimal state actions demonstrations are difficult for the teacher to provide. For example in robotics, providing kinesthetic demonstrations on a robotic manipulator requires the teacher to control multiple degrees of freedom at once. The difficulty of requiring optimal state action demonstrations limits the space of problems where the teacher can provide quality feedback. As an alternative to state action demonstrations, the teacher can provide corrective feedback such as their preferences or rewards. Prior work has created algorithms designe…
In many contemporary applications such as healthcare, finance, robotics, and recommendation systems, continuous deployment of new policies for data collection and online learning is either cost ineffective or impractical. We consider a setting that lies between pure offline reinforcement learning (RL) and pure online RL called deployment constrained RL in which the number of policy deployments for data sampling is limited. To solve this challenging task, we propose a new algorithmic learning framework called Model-based Uncertainty regularized and Sample Efficient Batch Optimization (MUSBO). O…
Mila, McGill University（麦吉尔大学米拉研究所）；Facebook（脸书）；Facebook AI Research（脸书人工智能研究院）
Widely-used deep reinforcement learning algorithms have been shown to fail in the batch setting--learning from a fixed data set without interaction with the environment. Following this result, there have been several papers showing reasonable performances under a variety of environments and batch settings. In this paper, we benchmark the performance of recent off-policy and batch reinforcement learning algorithms under unified settings on the Atari domain, with data generated by a single partially-trained behavioral policy. We find that under these conditions, many of these algorithms underper…
End-to-end learning robotic manipulation with high data efficiency is one of the key challenges in robotics. The latest methods that utilize human demonstration data and unsupervised representation learning has proven to be a promising direction to improve RL learning efficiency. The use of demonstration data also allows "warming-up" the RL policies using offline data with imitation learning or the recently emerged offline reinforcement learning algorithms. However, existing works often treat offline policy learning and online exploration as two separate processes, which are often accompanied …
We present a framework for data-driven robotics that makes use of a large dataset of recorded robot experience and scales to several tasks using learned reward functions. We show how to apply this framework to accomplish three different object manipulation tasks on a real robot platform. Given demonstrations of a task together with task-agnostic recorded experience, we use a special form of human annotation as supervision to learn a reward function, which enables us to deal with real-world tasks where the reward signal cannot be acquired directly. Learned rewards are used in combination with a…
Data generation and labeling are usually an expensive part of learning for robotics. While active learning methods are commonly used to tackle the former problem, preference-based learning is a concept that attempts to solve the latter by querying users with preference questions. In this paper, we will develop a new algorithm, batch active preference-based learning, that enables efficient learning of reward functions using as few data samples as possible while still having short query generation times. We introduce several approximations to the batch active learning problem, and provide theore…
Data driven robotics relies upon accurate real-world representations to learn useful policies. Despite our best-efforts, zero-shot sim-to-real transfer is still an unsolved problem, and we often need to allow our agents to explore online to learn useful policies for a given task. For many applications of field robotics online exploration is prohibitively expensive and dangerous, this is especially true in fixed-wing aerial robotics. To address these challenges we offer an intermediary solution for learning in field robotics. We investigate the use of dissimilar platform vehicle for learning an…
In this paper, we explore the problem of interesting scene prediction for mobile robots. This area is currently underexplored but is crucial for many practical applications such as autonomous exploration and decision making. Inspired by industrial demands, we first propose a novel translation-invariant visual memory for recalling and identifying interesting scenes, then design a three-stage architecture of long-term, short-term, and online learning. This enables our system to learn human-like experience, environmental knowledge, and online adaption, respectively. Our approach achieves much hig…
Queensland University of Technology (QUT)（昆士兰科技大学）；DeepMind（深度思维）
Deep reinforcement learning has been shown to solve challenging tasks where large amounts of training experience is available, usually obtained online while learning the task. Robotics is a significant potential application domain for many of these algorithms, but generating robot experience in the real world is expensive, especially when each task requires a lengthy online training procedure. Off-policy algorithms can in principle learn arbitrary tasks from a diverse enough fixed dataset. In this work, we evaluate popular exploration methods by generating robotics datasets for the purpose of …
University of Applied Sciences and Arts of Southern Switzerland (SUPSI)（瑞士南部应用科学与艺术大学）；Ecole Polytechnique Fédérale de Lausanne (EPFL)（洛桑联邦理工学院）
Temporary school closures caused by the Covid-19 pandemic have posed new challenges for many teachers and students worldwide. Especially the abrupt shift to online distance learning posed many obstacles to be overcome and it particularly complicated the implementation of Educational Robotics activities. Such activities usually comprise a variety of different learning artifacts, which were not accessible to many students during the period of school closure. Moreover, online distance learning considerably limits the possibilities for students to interact with their peers and teachers. In an atte…
The use of rehabilitation robotics in clinical applications gains increasing importance, due to therapeutic benefits and the ability to alleviate labor-intensive works. However, their practical utility is dependent on the deployment of appropriate control algorithms, which adapt the level of task-assistance according to each individual patient's need. Generally, the required personalization is achieved through manual tuning by clinicians, which is cumbersome and error-prone. In this work we propose a novel online learning control architecture, which is able to personalize the control force at …
Robotic Grasping of Harvested Tomato Trusses Using Vision and Online Learning
Delft University of Technology（代尔夫特理工大学）；Czech Technical University in Prague（布拉格捷克技术大学）；Czech Institute of Informatics, Robotics, and Cybernetics（捷克信息学、机器人学与控制论研究所）
Tsinghua University（清华大学）；The Chinese University of Hong Kong（香港中文大学）
The deformable linear objects (DLOs) are common in both industrial and domestic applications, such as wires, cables, ropes. Because of its highly deformable nature, it is difficult for the robot to reproduce human's dexterous skills on DLOs. In this paper, the unknown deformation model is estimated in both the offline and online manners. The offline learning aims to provide a good approximation prior to the manipulation task, while the online learning aims to compensate the errors due to insufficient training (e.g. limited datasets) in the offline phase. The offline module works by constructin…
We propose a solution for sensor extrinsic self-calibration with very low time complexity, competitive accuracy and graceful handling of often-avoided corner cases: drift in calibration parameters and unobservable directions in the parameter space. It consists of three main parts: 1) information-theoretic based segment selection for constant-time estimation; 2) observability-aware parameter update through a rank-revealing decomposition of the Fisher information matrix; 3) drift-correcting self-calibration through the time-decay of segments. At the core of our FastCal algorithm is the loosely-c…
In this paper, we propose an approach to designing online feedback controllers for input-saturated robotic systems evolving on Lie groups by extending the recently developed Sequential Action Control (SAC). In contrast to existing feedback controllers, our approach poses the nonconvex constrained nonlinear optimization problem as the tracking of a desired negative mode insertion gradient on the configuration space of a Lie group. This results in a closed-form feedback control law even with input saturation and thus is well suited for online application. In extending SAC to Lie groups, the asso…
Robotics Institute, University of Michigan（密歇根大学机器人研究所）；Department of Mechanical Engineering, University of Michigan（密歇根大学机械工程系）；Department of Naval Architecture and Marine Engineering, University of Michigan（密歇根大学 naval architecture and marine engineering系）
Online control design using a high-fidelity, full-order model for a bipedal robot can be challenging due to the size of the state space of the model. A commonly adopted solution to overcome this challenge is to approximate the full-order model (anchor) with a simplified, reduced-order model (template), while performing control synthesis. Unfortunately it is challenging to make formal guarantees about the safety of an anchor model using a controller designed in an online fashion using a template model. To address this problem, this paper proposes a method to generate safety-preserving controlle…
Can a robot be a personal dating coach? Even with the increasing amount of conversational data on the internet, the implementation of conversational robots remains a challenge. In particular, a detailed and professional counseling log is expensive and not publicly accessible. In this paper, we develop a robot dating coaching system leveraging corpus from online communities. We examine people's perceptions of the dating coaching robot with a dialogue module. 97 participants joined to have a conversation with the robot, and 30 of them evaluated the robot. The results indicate that participants t…
Robotics Institute, Carnegie Mellon University（卡内基梅隆大学机器人研究所）
Autonomous robots frequently need to detect "interesting" scenes to decide on further exploration, or to decide which data to share for cooperation. These scenarios often require fast deployment with little or no training data. Prior work considers "interestingness" based on data from the same distribution. Instead, we propose to develop a method that automatically adapts online to the environment to report interesting scenes quickly. To address this problem, we develop a novel translation-invariant visual memory and design a three-stage architecture for long-term, short-term, and online learn…
University of Tsukuba（筑波大学）；RIKEN Brain Science Institute（日本理化学研究所脑科学研究所）；The University of Tokyo（东京大学）
We report on an extended robot control application of a contact-less and airborne ultrasonic tactile display (AUTD) stimulus-based brain-computer interface (BCI) paradigm, which received last year The Annual BCI Research Award 2014. In the award winning human communication augmentation paradigm the six palm positions are used to evoke somatosensory brain responses, in order to define a novel contactless tactile BCI. An example application of a small robot management is also presented in which the users control a small robot online.
ONERA（法国航空航天实验室）；Université Paris-Saclay（巴黎萨克雷大学）；University of Toulouse（图卢兹大学）
This paper introduces an Online Localisation and Colored Mesh Reconstruction (OLCMR) ROS perception architecture for ground exploration robots aiming to perform robust Simultaneous Localisation And Mapping (SLAM) in challenging unknown environments and provide an associated colored 3D mesh representation in real time. It is intended to be used by a remote human operator to easily visualise the mapped environment during or after the mission or as a development base for further researches in the field of exploration robotics. The architecture is mainly composed of carefully-selected open-source …
RoboBPP: Benchmarking Robotic Online Bin Packing with Physics-based Simulation
National University of Defense Technology（国防科技大学）；Institute of AI for Industries, Chinese Academy of Sciences（中国科学院人工智能产业研究院）；Wuhan University（武汉大学）；Shenzhen University（深圳大学）；JD Explore Academy（京东探索研究院）；China Post Technology Company（中国邮政技术公司）
SGT Inc.（SGT公司）；NASA Ames Research Center（美国国家航空航天局艾姆斯研究中心）；Institute for Systems and Robotics（系统与机器人研究所）；Instituto Superior Técnico（里斯本高等理工学院）；Massachusetts Institute of Technology（麻省理工学院）
Space free-flyers like the Astrobee robots currently operating aboard the International Space Station must operate with inherent system uncertainties. Parametric uncertainties like mass and moment of inertia are especially important to quantify in these safety-critical space systems and can change in scenarios such as on-orbit cargo movement, where unknown grappled payloads significantly change the system dynamics. Cautiously learning these uncertainties en route can potentially avoid time- and fuel-consuming pure system identification maneuvers. Recognizing this, this work proposes RATTLE, an…
Fillet welding is one of the most widespread types of welding in the industry, which is still carried out manually or automated by contact. This paper aims to describe an online programming system for noncontact fillet welding robots with U and L shaped structures, which responds to the needs of the Fourth Industrial Revolution. In this paper, the authors propose an online robot programming methodology that eliminates unnecessary steps traditionally performed in robotic welding, so that the operator only performs three steps to complete the welding task. First, choose the piece to weld. Then, …
