1 code implementation • 11 Apr 2023 • Maximilian Xiling Li, Onur Celik, Philipp Becker, Denis Blessing, Rudolf Lioutikov, Gerhard Neumann
Learning skills by imitation is a promising concept for the intuitive teaching of robots.
no code implementations • 3 Apr 2023 • Niklas Freymuth, Philipp Dahlinger, Tobias Würth, Luise Kärger, Gerhard Neumann
Adaptive Mesh Refinement (AMR) is crucial for mesh-based simulations, as it allows for dynamically adjusting the resolution of a mesh to trade off computational cost with the simulation accuracy.
no code implementations • 27 Mar 2023 • Denis Blessing, Onur Celik, Xiaogang Jia, Moritz Reuss, Maximilian Xiling Li, Rudolf Lioutikov, Gerhard Neumann
We propose a novel curriculum-based approach to learning mixture models in which each component of the MoE is able to select its own subset of the training data for learning.
1 code implementation • 23 Feb 2023 • Jonas Linkerhägner, Niklas Freymuth, Paul Maria Scheikl, Franziska Mathis-Ullrich, Gerhard Neumann
Our method results in utilization of additional point cloud information to accurately predict stable simulations where existing Graph Network Simulators fail.
no code implementations • 10 Feb 2023 • Philipp Becker, Sebastian Markgraf, Fabian Otto, Gerhard Neumann
We demonstrate the benefits of utilizing proprioception in learning representations for RL on a large set of experiments.
no code implementations • 18 Oct 2022 • Fabian Otto, Onur Celik, Hongyi Zhou, Hanna Ziesche, Ngo Anh Vien, Gerhard Neumann
In this paper, we present a new algorithm for deep ERL.
1 code implementation • 17 Oct 2022 • Niklas Freymuth, Nicolas Schreiber, Philipp Becker, Aleksandar Taranovic, Gerhard Neumann
We find that the geometric descriptors greatly help in generalizing to new task configurations and that combining them with our distribution-matching objective is crucial for representing and reproducing versatile behavior.
1 code implementation • 17 Oct 2022 • Philipp Becker, Gerhard Neumann
We show that RSSMs use a suboptimal inference scheme and that models trained using this inference overestimate the aleatoric uncertainty of the ground truth system.
Model-based Reinforcement Learning reinforcement-learning +1
no code implementations • 4 Oct 2022 • Ge Li, Zeqi Jin, Michael Volpp, Fabian Otto, Rudolf Lioutikov, Gerhard Neumann
MPs can be broadly categorized into two types: (a) dynamics-based approaches that generate smooth trajectories from any initial state, e. g., Dynamic Movement Primitives (DMPs), and (b) probabilistic approaches that capture higher-order statistics of the motion, e. g., Probabilistic Movement Primitives (ProMPs).
no code implementations • 23 Sep 2022 • Oleg Arenz, Philipp Dahlinger, Zihan Ye, Michael Volpp, Gerhard Neumann
Variational inference with Gaussian mixture models (GMMs) enables learning of highly-tractable yet multi-modal approximations of intractable target distributions.
no code implementations • 22 Sep 2022 • Fabian Duffhauss, Ngo Anh Vien, Hanna Ziesche, Gerhard Neumann
Sensor fusion can significantly improve the performance of many computer vision tasks.
no code implementations • 1 Aug 2022 • Fabian Duffhauss, Tobias Demmler, Gerhard Neumann
We overcome this issue with our novel multi-view 6D pose estimation method called MV6D which accurately predicts the 6D poses of all objects in a cluttered scene based on RGB-D images from multiple perspectives.
no code implementations • 31 Jul 2022 • Abdalkarim Mohtasib, Gerhard Neumann, Heriberto Cuayahuitl
Learning robotic tasks in the real world is still highly challenging and effective practical solutions remain to be found.
1 code implementation • ICLR 2022 • Vaisakh Shaj, Dieter Buchler, Rohit Sonker, Philipp Becker, Gerhard Neumann
Recurrent State-space models (RSSMs) are highly expressive models for learning patterns in time series data and system identification.
no code implementations • 14 Jun 2022 • Yumeng Li, Ning Gao, Hanna Ziesche, Gerhard Neumann
We present a novel meta-learning approach for 6D pose estimation on unknown objects.
no code implementations • 27 May 2022 • Moritz Reuss, Niels van Duijkeren, Robert Krug, Philipp Becker, Vaisakh Shaj, Gerhard Neumann
These models need to precisely capture the robot dynamics, which consist of well-understood components, e. g., rigid body dynamics, and effects that remain challenging to capture, e. g., stick-slip friction and mechanical flexibilities.
no code implementations • 24 May 2022 • Maximilian Hüttenrauch, Gerhard Neumann
In contrast, stochastic optimizers that are motivated by policy gradients, such as the Model-based Relative Entropy Stochastic Search (MORE) algorithm, directly optimize the expected fitness function without the use of rankings.
no code implementations • 23 May 2022 • Ruijie Chen, Ning Gao, Ngo Anh Vien, Hanna Ziesche, Gerhard Neumann
Grasping inhomogeneous objects, practical use in real-world applications, remains a challenging task due to the unknown physical properties such as mass distribution and coefficient of friction.
no code implementations • 15 Mar 2022 • Hadi Beik-Mohammadi, Søren Hauberg, Georgios Arvanitidis, Gerhard Neumann, Leonel Rozo
We argue that Riemannian manifolds may be learned via human demonstrations in which geodesics are natural motion skills.
2 code implementations • CVPR 2022 • Ning Gao, Hanna Ziesche, Ngo Anh Vien, Michael Volpp, Gerhard Neumann
To this end, we (i) exhaustively evaluate common meta-learning techniques on these tasks, and (ii) quantitatively analyze the effect of various deep learning techniques commonly used in recent meta-learning algorithms in order to strengthen the generalization capability: data augmentation, domain randomization, task augmentation and meta-regularization.
1 code implementation • 8 Dec 2021 • Onur Celik, Dongzhuoran Zhou, Ge Li, Philipp Becker, Gerhard Neumann
This local and incremental learning results in a modular MoE model of high accuracy and versatility, where both properties can be scaled by adding more components on the fly.
no code implementations • 16 Nov 2021 • Giao Nguyen-Quynh, Philipp Becker, Chen Qiu, Maja Rudolph, Gerhard Neumann
In addition, driving data can often be multimodal in distribution, meaning that there are distinct predictions that are likely, but averaging can hurt model performance.
no code implementations • 15 Nov 2021 • Niklas Freymuth, Philipp Becker, Gerhard Neumann
Inverse Reinforcement Learning infers a reward function from expert demonstrations, aiming to encode the behavior and intentions of the expert.
no code implementations • 29 Sep 2021 • Oleg Arenz, Zihan Ye, Philipp Dahlinger, Gerhard Neumann
Effective approaches for Gaussian variational inference are MORE, VOGN, and VON, which are zero-order, first-order, and second-order, respectively.
no code implementations • 6 Aug 2021 • Abdalkarim Mohtasib, Gerhard Neumann, Heriberto Cuayahuitl
We argue that it is crucial to automate the reward learning process so that new skills can be taught to robots by their users.
no code implementations • 10 Jun 2021 • Ngo Anh Vien, Gerhard Neumann
This paper proposes a differentiable robust LQR layer for reinforcement learning and imitation learning under model uncertainty and stochastic dynamics.
no code implementations • 8 Jun 2021 • Alireza Ranjbar, Ngo Anh Vien, Hanna Ziesche, Joschka Boedecker, Gerhard Neumann
We propose a new formulation that addresses these limitations by also modifying the feedback signals to the controller with an RL policy and show superior performance of our approach on a contact-rich peg-insertion task under position and orientation uncertainty.
no code implementations • 8 Jun 2021 • Hadi Beik-Mohammadi, Søren Hauberg, Georgios Arvanitidis, Gerhard Neumann, Leonel Rozo
For robots to work alongside humans and perform in unstructured environments, they must learn new motion skills and adapt them to unseen situations on the fly.
1 code implementation • ICLR 2021 • Fabian Otto, Philipp Becker, Ngo Anh Vien, Hanna Carolin Ziesche, Gerhard Neumann
However, enforcing such trust regions in deep reinforcement learning is difficult.
no code implementations • ICLR 2021 • Michael Volpp, Fabian Flürenbrock, Lukas Grossberger, Christian Daniel, Gerhard Neumann
Recently, casting probabilistic regression as a multi-task learning problem in terms of conditional latent variable (CLV) models such as the Neural Process (NP) has shown promising results.
1 code implementation • 20 Oct 2020 • Vaisakh Shaj, Philipp Becker, Dieter Buchler, Harit Pandya, Niels van Duijkeren, C. James Taylor, Marc Hanheide, Gerhard Neumann
We adopt a recent probabilistic recurrent neural network architecture, called Re-current Kalman Networks (RKNs), to model learning by conditioning its transition dynamics on the control actions.
no code implementations • 10 Aug 2020 • RB Ashith Shyam, Zhou Hao, Umberto Montanaro, Gerhard Neumann
Since actual hardware implementation of microgravity environment is extremely expensive, the demonstration data for trajectory learning is generated using a model predictive controller (MPC) in a physics based simulator.
1 code implementation • 8 Aug 2020 • Oleg Arenz, Gerhard Neumann
We also show that our non-adversarial formulation can be used to derive novel algorithms by presenting a method for offline imitation learning that is inspired by the recent ValueDice algorithm, but does not rely on small policy updates for convergence.
1 code implementation • ICLR 2020 • Philipp Becker, Oleg Arenz, Gerhard Neumann
Such behavior is appealing whenever we deal with highly multi-modal data where modelling single modes correctly is more important than covering all the modes.
no code implementations • 10 Jul 2019 • Oleg Arenz, Mingjun Zhong, Gerhard Neumann
For efficient improvement of the GMM approximation, we derive a lower bound on the corresponding optimization objective enabling us to update the components independently.
3 code implementations • 17 May 2019 • Philipp Becker, Harit Pandya, Gregor Gebhardt, Cheng Zhao, James Taylor, Gerhard Neumann
In order to integrate uncertainty estimates into deep time-series modelling, Kalman Filters (KFs) (Kalman et al., 1960) have been integrated with deep learning models, however, such approaches typically rely on approximate inference techniques such as variational inference which makes learning more complex and often less scalable due to approximation errors.
no code implementations • 7 Feb 2019 • Joni Pajarinen, Hong Linh Thai, Riad Akrour, Jan Peters, Gerhard Neumann
Trust-region methods have yielded state-of-the-art results in policy search.
no code implementations • 16 Nov 2018 • Takayuki Osa, Joni Pajarinen, Gerhard Neumann, J. Andrew Bagnell, Pieter Abbeel, Jan Peters
This process of learning from demonstrations, and the study of algorithms to do so, is called imitation learning.
1 code implementation • 31 Aug 2018 • Sebastian Gomez-Gonzalez, Gerhard Neumann, Bernhard Schölkopf, Jan Peters
However, to be able to capture variability and correlations between different joints, a probabilistic movement primitive requires the estimation of a larger number of parameters compared to their deterministic counterparts, that focus on modeling only the mean behavior.
no code implementations • 20 Aug 2018 • Patrick Jahnke, Emmanuel Stapf, Jonas Mieseler, Gerhard Neumann, Patrick Eugster
In this space, into which we transform the input data via a Short-Time Fourier Transform (STFT), the peak structures of flows can be predicted after gleaning their key characteristics, with a Principal Component Analysis (PCA), from past and ongoing flows that stem from the same socket-to-socket connection.
1 code implementation • 17 Jul 2018 • Maximilian Hüttenrauch, Adrian Šošić, Gerhard Neumann
However, concatenation scales poorly to swarm systems with a large number of homogeneous agents as it does not exploit the fundamental properties inherent to these systems: (i) the agents in the swarm are interchangeable and (ii) the exact number of agents in the swarm is irrelevant.
1 code implementation • ICML 2018 • Oleg Arenz, Gerhard Neumann, Mingjun Zhong
Inference from complex distributions is a common problem in machine learning needed for many Bayesian methods.
no code implementations • 21 Sep 2017 • Maximilian Hüttenrauch, Adrian Šošić, Gerhard Neumann
Swarm systems constitute a challenging problem for reinforcement learning (RL) as the algorithm needs to learn decentralized control policies that can cope with limited local sensing and communication abilities of the agents.
1 code implementation • 18 Sep 2017 • Maximilian Hüttenrauch, Adrian Šošić, Gerhard Neumann
Here, we follow a guided approach where a critic has central access to the global state during learning, which simplifies the policy evaluation problem from a reinforcement learning point of view.
no code implementations • ICML 2017 • Riad Akrour, Dmitry Sorokin, Jan Peters, Gerhard Neumann
Bayesian optimization is renowned for its sample efficiency but its application to higher dimensional tasks is impeded by its focus on global optimization.
no code implementations • NeurIPS 2016 • Boris Belousov, Gerhard Neumann, Constantin A. Rothkopf, Jan R. Peters
In this paper, we show that interception strategies appearing to be heuristics can be understood as computational solutions to the optimal control problem faced by a ball-catching agent acting under uncertainty.
no code implementations • 10 Nov 2016 • Voot Tangkaratt, Herke van Hoof, Simone Parisi, Gerhard Neumann, Jan Peters, Masashi Sugiyama
A naive application of unsupervised dimensionality reduction methods to the context variables, such as principal component analysis, is insufficient as task-relevant input may be ignored.
no code implementations • 29 Jun 2016 • Riad Akrour, Abbas Abdolmaleki, Hany Abdulsamad, Jan Peters, Gerhard Neumann
In order to show the monotonic improvement of our algorithm, we additionally conduct a theoretical analysis of our policy update scheme to derive a lower bound of the change in policy return between successive iterations.
no code implementations • NeurIPS 2015 • Abbas Abdolmaleki, Rudolf Lioutikov, Jan R. Peters, Nuno Lau, Luis Pualo Reis, Gerhard Neumann
Stochastic search algorithms are general black-box optimizers.
no code implementations • NeurIPS 2013 • Alexandros Paraschos, Christian Daniel, Jan R. Peters, Gerhard Neumann
In order to use such a trajectory distribution for robot movement control, we analytically derive a stochastic feedback controller which reproduces the given trajectory distribution.
no code implementations • NeurIPS 2008 • Gerhard Neumann, Jan R. Peters
Recently, fitted Q-iteration (FQI) based methods have become more popular due to their increased sample efficiency, a more stable learning process and the higher quality of the resulting policy.