检索结果-内蒙古大学图书馆

Σ-Optimality for Active learning on Gaussian Random Fields 13

学校读者我要写书评

暂无评论

Σ-Optimality for Active Learning on Gaussian Random Fields

Annual Conference on Neural Information Processing Systems

作者： Yifei Ma Roman Garnett Jeff Schneider Machine Learning Department Carnegie Mellon University Computer Science Department University of Bonn Robotics Institute Carnegie Mellon University

ISBN: (纸本)9781632660244

A common classifier for unlabeled nodes on undirected graphs uses label propagation from the labeled nodes, equivalent to the harmonic predictor on Gaussian random fields (grfs). For active learning on grfs, the commonly used V-optimality criterion queries nodes that reduce the L~2 (regression) loss. V-optimality satisfies a submodularity property showing that greedy reduction produces a (1 - 1/e) globally optimal solution. However, L~2 loss may not characterise the true nature of 0/1 loss in classification problems and thus may not be the best choice for active learning. We consider a new criterion we call Σ-optimality, which queries the node that minimizes the sum of the elements in the predictive covariance. Σ-optimality directly optimizes the risk of the surveying problem, which is to determine the proportion of nodes belonging to one class. In this paper we extend submodularity guarantees from V-optimality to Σ-optimality using properties specific to grfs. We further show that GRFs satisfy the suppressor-free condition in addition to the conditional independence inherited from Markov random fields. We test Σ-optimality on real-world graphs with both synthetic and real data and show that it outperforms V-optimality and other related methods on classification.

关键词： Experiential learning Markov random fields Ganglia Conditional Independence Gauss Taxonomy Fields

One Policy to Run Them All: an End-to-end learning Approach to Multi-Embodiment Locomotion 8

学校读者我要写书评

暂无评论

One Policy to Run Them All: an End-to-end Learning Approach ...

8th Conference on Robot learning, CoRL 2024

作者： Bohlinger, Nico Czechmanowski, Grzegorz Krupka, Maciej Kicki, Piotr Walas, Krzysztof Peters, Jan Tateo, Davide Department of Computer Science Technical University of Darmstadt Germany Institute of Robotics and Machine Intelligence Poznan University of Technology Poland Research Department: Systems AI for Robot Learning Germany IDEAS NCBR Warsaw Poland Hessian.AI Germany Centre for Cognitive Science Germany

Deep Reinforcement learning techniques are achieving state-of-the-art results in robust legged locomotion. While there exists a wide variety of legged platforms such as quadruped, humanoids, and hexapods, the field is still missing a single learning framework that can control all these different embodiments easily and effectively and possibly transfer, zero or few-shot, to unseen robot embodiments. We introduce URMA, the Unified Robot Morphology Architecture, to close this gap. Our framework brings the end-to-end Multi-Task Reinforcement learning approach to the realm of legged robots, enabling the learned policy to control any type of robot morphology. The key idea of our method is to allow the network to learn an abstract locomotion controller that can be seamlessly shared between embodiments thanks to our morphology-agnostic encoders and decoders. This flexible architecture can be seen as a potential first step in building a foundation model for legged robot locomotion. Our experiments show that URMA can learn a locomotion policy on multiple embodiments that can be easily transferred to unseen robot platforms in simulation and the real world. © 2024 Proceedings of machine learning Research.

关键词： Deep reinforcement learning

BATS: Best Action Trajectory Stitching

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Char, Ian Mehta, Viraj Villaflor, Adam Dolan, John M. Schneider, Jeff Department of Machine Learning Carnegie Mellon University United States Robotics Institute Carnegie Mellon University United States

The problem of offline reinforcement learning focuses on learning a good policy from a log of environment interactions. Past efforts for developing algorithms in this area have revolved around introducing constraints to online reinforcement learning algorithms to ensure the actions of the learned policy are constrained to the logged data. In this work, we explore an alternative approach by planning on the fixed dataset directly. Specifically, we introduce an algorithm which forms a tabular Markov Decision Process (MDP) over the logged data by adding new transitions to the dataset. We do this by using learned dynamics models to plan short trajectories between states. Since exact value iteration can be performed on this constructed MDP, it becomes easy to identify which trajectories are advantageous to add to the MDP. Crucially, since most transitions in this MDP come from the logged data, trajectories from the MDP can be rolled out for long periods with confidence. We prove that this property allows one to make upper and lower bounds on the value function up to appropriate distance metrics. Finally, we demonstrate empirically how algorithms that uniformly constrain the learned policy to the entire dataset can result in unwanted behavior, and we show an example in which simply behavior cloning the optimal policy of the MDP created by our algorithm avoids this problem. Copyright © 2022, The Authors. All rights reserved.

关键词： Markov processes

Opening a lockbox through physical exploration

学校读者我要写书评

暂无评论

Opening a lockbox through physical exploration

IEEE-RAS International Conference on Humanoid Robots

作者： Manuel Baum Matthew Bernstein Roberto Martin-Martin Sebastian Höfer Johannes Kulick Marc Toussaint Alex Kacelnik Oliver Brock Robotics and Biology Lab (RBO) Technische Universität Berlin Machine Learning and Robotics Lab (MLR) Universität Stuttgart Department of Zoology Oxford University

ISBN: (纸本)9781538646793

How can we close the gap between animals and robots when it comes to intelligently interacting with the environment? On our quest for answers, we have investigated the problem of physically exploring complex mechanical puzzles, called lockboxes. Biologists have discovered that cockatoos are intrinsically motivated to explore and solve such problems through physical explorative behavior. In this work, we study how different strategies shape the robots' exploration, given basic perception-action skills. Our evaluation highlights the influence of different statistical priors on the performance of the exploration strategies, showing that not only a range of computational methods, but also a range of priors could explain different exploration behaviors. We carry out our study of exploration strategies both in simulation and on two robot platforms. This first step towards a fully integrated real-world system allowed us to identify and remove limitations of our prior theoretical work on cross-entropy-based exploration when applied to complex realistic scenarios. In this paper we propose novel variants of this strategy and our experiments verify that the cross-entropy method performs well on a physical lockbox analogue of the cockatoo apparatus, and can generalize to lockboxes of different properties.

关键词： Robot sensing systems Entropy Kinematics Robot kinematics Planning

Inferring capabilities by experimentation

学校读者我要写书评

暂无评论

arXiv 2017年

作者： Khadke, Ashwin Veloso, Manuela Robotics Institute Carnegie Mellon University United States Department of Machine Learning Carnegie Mellon University United States

We present an approach to enable an autonomous agent (learner) in building a model of a new unknown robot's (subject) performance at a task through experimentation. The subject's appearance can provide cues to its physical as well as cognitive capabilities. Building on these cues, our active experimentation approach learns a model that captures the effect of relevant extrinsic factors on the subject's ability to perform a task. As personal robots become increasingly multi-functional and adaptive, such autonomous agents would find use as tools for humans in determining"What can this robot do?". We applied our algorithm in modelling a NAO and a Pepper robot at two different tasks. We first demonstrate the advantages of our active experimentation approach, then we show the utility of such models in identifying scenarios a robot is well suited for, in performing a task. Copyright © 2017, The Authors. All rights reserved.

关键词： Autonomous agents

TReR: A Lightweight Transformer Re-Ranking Approach for 3D LiDAR Place Recognition

学校读者我要写书评

暂无评论

TReR: A Lightweight Transformer Re-Ranking Approach for 3D L...

International Conference on Intelligent Transportation

作者： Tiago Barros Luís Garrote Martin Aleksandrov Cristiano Premebida Urbano J. Nunes Department of Electrical and Computer Engineering University of Coimbra Institute of Systems and Robotics Portugal Dahlem Center for Machine Learning and Robotics Freie Universität Berlin Berlin

Autonomous driving systems often require reliable loop closure detection to guarantee reduced localization drift. Recently, 3D LiDAR-based localization methods have used retrieval-based place recognition to find revisited places efficiently. However, when deployed in challenging real-world scenarios, the place recognition models become more complex, which comes at the cost of high computational demand. This work tackles this problem from an information-retrieval perspective, adopting a first-retrieve-then-re-ranking paradigm, where an initial loop candidate ranking, generated from a 3D place recognition model, is re-ordered by a proposed lightweight transformer-based re-ranking approach (TReR). The proposed approach relies on global descriptors only, being agnostic to the place recognition model. The experimental evaluation, conducted on the KITTI Odometry dataset, where we compared TReR with s.o.t.a. re-ranking approaches such as αQE and SGV, indicate the robustness and efficiency when compared to αQE while offering a good trade-off between robustness and efficiency when compared to SGV.

关键词：

Long-Horizon Multi-Robot Rearrangement Planning for Construction Assembly

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Hartmann, Valentin N. Orthey, Andreas Driess, Danny Oguz, Ozgur S. Toussaint, Marc Machine Learning & Robotics Lab University of Stuttgart Germany Learning and Intelligent Systems Group TU Berlin Germany Department of Computer Engineering Bilkent University Turkey

Robotic assembly planning enables architects to explicitly account for the assembly process during the design phase, and enables efficient building methods that profit from the robots' different capabilities. Previous work has addressed planning of robot assembly sequences and identifying the feasibility of architectural designs. This paper extends previous work by enabling planning with large, heterogeneous teams of robots. We present a planning system which enables parallelization of complex task and motion planning problems by iteratively solving smaller subproblems. Combining optimization methods to solve for manipulation constraints with a sampling-based bi-directional space-time path planner enables us to plan cooperative multirobot manipulation with unknown arrival-times. Thus, our solver allows for completing subproblems and tasks with differing timescales and synchronizes them effectively. We demonstrate the approach on multiple case-studies to show the robustness over long planning horizons and scalability to many objects and agents of our algorithm. Finally, we also demonstrate the execution of the computed plans on two robot arms to showcase the feasibility in the real world. Copyright © 2021, The Authors. All rights reserved.

关键词： robotics

Fast planning for dynamic preferences

学校读者我要写书评

暂无评论

Fast planning for dynamic preferences

18th International Conference on Automated Planning and Scheduling, ICAPS 2008

作者： Ziebart, Brian D. Dey, Anind K. Bagnell, J. Andrew Machine Learning Department Carnegie Mellon University Pittsburgh PA 15213 United States Human-Computer Interaction Institute Carnegie Mellon University Pittsburgh PA 15213 United States Robotics Institute Carnegie Mellon University Pittsburgh PA 15213 United States

ISBN: (纸本)9781577353867

We present an algorithm that quickly finds optimal plans for unforeseen agent preferences within graph-based planning domains where actions have deterministic outcomes and action costs are linearly parameterized by preference parameters. We focus on vehicle route planning for drivers with personal trade-offs for different types of roads, and specifically on settings where these preferences are not known until planning time. We employ novel bounds (based on the triangle inequality and on the the concavity of the optimal plan cost in the space of preferences) to enable the reuse of previously computed optimal plans that are similar to the new plan preferences. The resulting lower bounds are employed to guide the search for the optimal plan up to 60 times more efficiently than previous methods. Copyright © 2008, Association for the Advancement of Artificial Intelligence (***). All rights reserved.

关键词： Graphic methods

TReR: A Lightweight Transformer Re-Ranking Approach for 3D LiDAR Place Recognition

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Barros, Tiago Garrote, Luís Aleksandrov, Martin Premebida, Cristiano Nunes, Urbano J. The University of Coimbra Institute of Systems and Robotics Department of Electrical and Computer Engineering Portugal Dahlem Center for Machine Learning and Robotics Freie Universität Berlin Berlin Germany

关键词： Efficiency