no code implementations • 28 Apr 2025 • Alexander Gräfe, Sebastian Trimpe
Yet, this central computer typically serves multiple systems simultaneously, leading to significant hardware demands due to the need to solve numerous optimization problems concurrently.
1 code implementation • 28 Jan 2025 • Bernd Frauenknecht, Devdutt Subhasish, Friedrich Solowjow, Sebastian Trimpe
Model-based reinforcement learning (MBRL) seeks to enhance data efficiency by learning a model of the environment and generating synthetic rollouts from it.
no code implementations • 23 Jan 2025 • Christian Fiedler, Johanna Menn, Sebastian Trimpe
A recurring and important task in control engineering is parameter tuning under constraints, which conceptually amounts to optimization of a blackbox function accessible only through noisy evaluations.
no code implementations • 22 Jan 2025 • Johanna Menn, Pietro Pelizzari, Michael Fleps-Dezasse, Sebastian Trimpe
To obtain safety guarantees, many existing safe Bayesian optimization methods rely on assumptions that are hard to verify in practice.
1 code implementation • 20 Jan 2025 • David Stenger, Dominik Scheurenberg, Heike Vallery, Sebastian Trimpe
Standard BO practice is to evaluate the closed-loop performance of parameters proposed during optimization on an episode with a fixed length.
1 code implementation • 14 Jan 2025 • Emma Cramer, Lukas Jäschke, Sebastian Trimpe
These results highlight the potential of adaptive hybrid RL for real-world, contact-rich tasks trained directly on hardware.
no code implementations • 12 Dec 2024 • Paul Brunzema, Mikkel Jordahn, John Willes, Sebastian Trimpe, Jasper Snoek, James Harrison
While Bayesian neural networks (BNNs) are a promising direction for higher capacity surrogate models, they have so far seen limited use due to poor performance on some problem types.
no code implementations • 30 Nov 2024 • Martin Ziegler, Andres Felipe Posada-Moreno, Friedrich Solowjow, Sebastian Trimpe
Our results demonstrate the feasibility of foundation models for dynamical systems that outperform specialist models in terms of generalization, data efficiency, and robustness.
no code implementations • 25 Nov 2024 • Alexander von Rohr, David Stenger, Dominik Scheurenberg, Sebastian Trimpe
Controller tuning is crucial for closed-loop performance but often involves manual adjustments.
1 code implementation • 21 Nov 2024 • Shiming He, Alexander von Rohr, Dominik Baumann, Ji Xiang, Sebastian Trimpe
At the core of the algorithm, a probabilistic model learns the dependence of the policy parameters and the robot learning objective not only by performing experiments on the robot, but also by leveraging data from a simulator.
no code implementations • 25 Sep 2024 • Leon Greiser, Ozan Demir, Benjamin Hartmann, Henrik Hose, Sebastian Trimpe
Instead, in a data-driven approach, recorded trajectories from the real system can be used to train local model networks (LMNs), for which feedforward controllers are derived via feedback linearization.
1 code implementation • 28 Jun 2024 • Emma Cramer, Bernd Frauenknecht, Ramil Sabirov, Sebastian Trimpe
Combining Reinforcement Learning (RL) with a prior controller can yield the best out of two worlds: RL can solve complex nonlinear problems, while the control prior ensures safer exploration and speeds up training.
1 code implementation • 26 Jun 2024 • Julian Dierkes, Emma Cramer, Holger H. Hoos, Sebastian Trimpe
We then propose a methodology for the combined optimisation of hyperparameters and the reward function.
no code implementations • 10 Jun 2024 • Pierre-François Massiani, Sebastian Trimpe, Friedrich Solowjow
We propose the new notion of empirical weak convergence (EWC) as a general assumption explaining such phenomena for kernel methods.
1 code implementation • 29 May 2024 • Bernd Frauenknecht, Artur Eisele, Devdutt Subhasish, Friedrich Solowjow, Sebastian Trimpe
This combination raises a critical question: 'When to trust your model?
no code implementations • 17 May 2024 • Guner Dilsad Er, Sebastian Trimpe, Michael Muehlebach
We consider a distributed learning problem, where agents minimize a global objective function by exchanging information over a network.
1 code implementation • 8 Apr 2024 • Henrik Hose, Alexander Gräfe, Sebastian Trimpe
By incorporating local sensitivities of nonlinear programs, the proposed method not only mimics optimal MPC inputs but also adjusts to known changes in physical parameters of the model using linear predictions while still guaranteeing stability.
no code implementations • 19 Mar 2024 • Christian Fiedler, Johanna Menn, Lukas Kreisköther, Sebastian Trimpe
To overcome this challenge, we introduce the Lipschitz-only Safe Bayesian Optimization (LoSBO) algorithm, which guarantees safety without an assumption on the RKHS bound, and empirically show that this algorithm is not only safe, but also exhibits superior performance compared to the state-of-the-art on several function classes.
1 code implementation • 15 Dec 2023 • Abdullah Tokmak, Christian Fiedler, Melanie N. Zeilinger, Sebastian Trimpe, Johannes Köhler
We address this problem by presenting a novel algorithm that automatically computes an explicit approximation to nonlinear MPC schemes while retaining closed-loop guarantees.
1 code implementation • 13 Dec 2023 • Antonia Holzapfel, Paul Brunzema, Sebastian Trimpe
Utilizing standard safe BO strategies that do not address time-variations can result in failure as previous safe decisions may become unsafe over time, which we demonstrate herein.
no code implementations • 11 Dec 2023 • Christian Fiedler, Michael Herty, Sebastian Trimpe
Mean field limits are an important tool in the context of large-scale dynamical systems, in particular, when studying multiagent and interacting particle systems.
1 code implementation • 1 Dec 2023 • Emma Cramer, Jonas Reiher, Sebastian Trimpe
However, whether an SAE is actually able to track objects in the scene and thus yields a spatial state representation well suited for RL tasks has rarely been examined due to a lack of established metrics.
no code implementations • 30 Nov 2023 • Bernd Frauenknecht, Tobias Ehlgen, Sebastian Trimpe
We find that in the case of trajectory control, the standard model-based RL formulation used in approaches like PETS-MPPI and MBPO is not suitable.
no code implementations • 27 Oct 2023 • Christian Fiedler, Michael Herty, Sebastian Trimpe
In many applications of machine learning, a large number of variables are considered.
no code implementations • 6 Sep 2023 • Katharina Ensinger, Sebastian Ziesche, Sebastian Trimpe
In this paper, we propose a new approach to hybrid modeling, where we inform the latent states of a learned model via a black-box simulator.
no code implementations • 5 Sep 2023 • Katharina Ensinger, Nicholas Tagliapietra, Sebastian Ziesche, Sebastian Trimpe
This is crucial to sample consistent predictions from the dynamics model.
1 code implementation • 11 Aug 2023 • Andrés Felipe Posada-Moreno, Lukas Kreisköther, Tassilo Glander, Sebastian Trimpe
Convolutional Neural Networks (CNN) have become a common choice for industrial quality control, as well as other critical applications in the Industry 4. 0.
no code implementations • 16 Jul 2023 • Shamil Mamedov, A. René Geist, Jan Swevers, Sebastian Trimpe
Accurately predicting deformable linear object (DLO) dynamics is challenging, especially when the task requires a model that is both human-interpretable and computationally efficient.
no code implementations • 6 Jun 2023 • Andrés Felipe Posada-Moreno, Kai Müller, Florian Brillowski, Friedrich Solowjow, Thomas Gries, Sebastian Trimpe
Subsequently, we demonstrate the potential of CE methods, by applying them in three industrial use cases.
no code implementations • 15 May 2023 • Lukas Kesper, Sebastian Trimpe, Dominik Baumann
Model-free learning of communication and control policies provides an alternative.
Multi-agent Reinforcement Learning
reinforcement-learning
+1
1 code implementation • 19 Apr 2023 • Henrik Hose, Johannes Köhler, Melanie N. Zeilinger, Sebastian Trimpe
Our method requires a single evaluation of the NN and forward integration of the input sequence online, which is fast to compute on resource-constrained systems.
1 code implementation • 8 Mar 2023 • Behnam Khojasteh, Friedrich Solowjow, Sebastian Trimpe, Katherine J. Kuchenbecker
Machine learning and deep learning have been used extensively to classify physical surfaces through images and time-series contact data.
no code implementations • 28 Feb 2023 • Christian Fiedler, Michael Herty, Michael Rom, Chiara Segala, Sebastian Trimpe
Kernel methods, being supported by a well-developed theory and coming with efficient algorithms, are among the most popular and successful machine learning techniques.
no code implementations • 27 Feb 2023 • Katharina Ensinger, Sebastian Ziesche, Barbara Rakitsch, Michael Tiemann, Sebastian Trimpe
This filtering technique combines two signals by applying a high-pass filter to one signal, and low-pass filtering the other.
1 code implementation • 23 Feb 2023 • Pierre-François Massiani, Mona Buisson-Fenet, Friedrich Solowjow, Florent Di Meglio, Sebastian Trimpe
Establishing these properties is challenging, especially when no analytical model is available and they are to be inferred directly from measurement data.
no code implementations • 30 Sep 2022 • Alexander Gräfe, Dominik Baumann, Sebastian Trimpe
We propose a fault detection method that uses these priorities to detect errors in other agents.
1 code implementation • 23 Aug 2022 • Paul Brunzema, Alexander von Rohr, Friedrich Solowjow, Sebastian Trimpe
To cope with stale data arising from time variations, current approaches to TVBO require prior knowledge of a constant rate of change.
1 code implementation • 28 Jul 2022 • Alexander von Rohr, Friedrich Solowjow, Sebastian Trimpe
However, in practice, many systems also exhibit uncertainty in the form of changes over time, e. g., due to weight shifts or wear and tear, leading to decreased performance or instability of the learning-based controller.
2 code implementations • 22 Jul 2022 • Paul Brunzema, Alexander von Rohr, Sebastian Trimpe
Current TVBO methods do not explicitly account for these properties, resulting in poor tuning performance and many unstable controllers through over-exploration of the parameter space.
1 code implementation • 22 Jun 2022 • Alexander Gräfe, Joram Eickhoff, Sebastian Trimpe
Distributed model predictive control (DMPC) is often used to tackle path planning for unmanned aerial vehicle (UAV) swarms.
no code implementations • 9 Jun 2022 • Andres Felipe Posada-Moreno, Nikita Surya, Sebastian Trimpe
Further, we introduce a process for the validation of concept-extraction techniques based on synthetic datasets with pixel-wise annotations of their main components, reducing the need for human intervention.
no code implementations • 25 May 2022 • Mona Buisson-Fenet, Valery Morgenthaler, Sebastian Trimpe, Florent Di Meglio
Identifying dynamical systems from experimental data is a notably difficult task.
no code implementations • 5 Apr 2022 • Sebastian Schlor, Friedrich Solowjow, Sebastian Trimpe
However, they usually come with a critical assumption - access to an accurate model of the system.
1 code implementation • 12 Feb 2022 • Sebastian Weichwald, Søren Wengel Mogensen, Tabitha Edith Lee, Dominik Baumann, Oliver Kroemer, Isabelle Guyon, Sebastian Trimpe, Jonas Peters, Niklas Pfister
Questions in causality, control, and reinforcement learning go beyond the classical machine learning task of prediction under i. i. d.
Model-based Reinforcement Learning
reinforcement-learning
+1
1 code implementation • 24 Jan 2022 • Bhavya Sukhija, Matteo Turchetta, David Lindner, Andreas Krause, Sebastian Trimpe, Dominik Baumann
Learning optimal control policies directly on physical systems is challenging since even a single failure can lead to costly hardware damage.
3 code implementations • NeurIPS 2021 • Sarah Müller, Alexander von Rohr, Sebastian Trimpe
We develop an algorithm utilizing a probabilistic model of the objective function and its gradient.
1 code implementation • 27 May 2021 • Dominik Baumann, Alonso Marco, Matteo Turchetta, Sebastian Trimpe
When learning policies for robotic systems from data, safety is a major concern, as violation of safety constraints may cause hardware damage.
1 code implementation • 25 May 2021 • Pierre-François Massiani, Steve Heim, Friedrich Solowjow, Sebastian Trimpe
Although it is often not possible to compute the minimum required penalty, we reveal clear structure of how the penalty, rewards, discount factor, and dynamics interact.
1 code implementation • 17 May 2021 • Pierre-François Massiani, Steve Heim, Sebastian Trimpe
In particular, we discuss the family of constraints that enforce safety in the context of a nominal control policy, and expose that these constraints do not need to be accurate everywhere.
1 code implementation • 17 May 2021 • Alexander von Rohr, Matthias Neumann-Brosig, Sebastian Trimpe
Probabilistic models such as Gaussian processes (GPs) are powerful tools to learn unknown dynamical systems from data for subsequent use in control design.
no code implementations • 7 May 2021 • Christian Fiedler, Carsten W. Scherer, Sebastian Trimpe
The combination of machine learning with control offers many opportunities, in particular for robust control.
1 code implementation • 6 May 2021 • Christian Fiedler, Carsten W. Scherer, Sebastian Trimpe
However, these estimates are of a Bayesian nature, whereas for some important applications, like learning-based control with safety guarantees, frequentist uncertainty bounds are required.
no code implementations • 16 Apr 2021 • Fabian Mager, Dominik Baumann, Carsten Herrmann, Sebastian Trimpe, Marco Zimmerling
An important class of cyber-physical systems relies on multiple agents that jointly perform a task by coordinating their actions over a wireless network.
no code implementations • 2 Feb 2021 • Katharina Ensinger, Friedrich Solowjow, Sebastian Ziesche, Michael Tiemann, Sebastian Trimpe
On the other hand, classical numerical integrators are specifically designed to preserve these crucial properties through time.
no code implementations • 11 Dec 2020 • A. René Geist, Sebastian Trimpe
Accurate models of mechanical system dynamics are often critical for model-based control and reinforcement learning.
1 code implementation • 16 Oct 2020 • Alonso Marco, Dominik Baumann, Majid Khadiv, Philipp Hennig, Ludovic Righetti, Sebastian Trimpe
We consider failing behaviors as those that violate a constraint and address the problem of learning with crash constraints, where no data is obtained upon constraint violation.
1 code implementation • 11 Aug 2020 • Niklas Funk, Dominik Baumann, Vincent Berenz, Sebastian Trimpe
We present a framework for model-free learning of event-triggered control strategies.
1 code implementation • L4DC 2020 • Andreas Geist, Sebastian Trimpe
The identification of the constrained dynamics of mechanical systems is often challenging.
1 code implementation • 6 Jun 2020 • Dominik Baumann, Friedrich Solowjow, Karl H. Johansson, Sebastian Trimpe
In this paper, we propose a method that identifies the causal structure of control systems.
1 code implementation • 15 May 2020 • Alonso Marco, Alexander von Rohr, Dominik Baumann, José Miguel Hernández-Lobato, Sebastian Trimpe
When learning to ride a bike, a child falls down a number of times before achieving the first success.
1 code implementation • 23 Apr 2020 • A. Rene Geist, Sebastian Trimpe
The identification of the constrained dynamics of mechanical systems is often challenging.
no code implementations • 23 Apr 2020 • Friedrich Solowjow, Dominik Baumann, Christian Fiedler, Andreas Jocham, Thomas Seel, Sebastian Trimpe
Evaluating whether data streams are drawn from the same distribution is at the heart of various machine learning problems.
no code implementations • 22 Dec 2019 • Julian Nubert, Johannes Köhler, Vincent Berenz, Frank Allgöwer, Sebastian Trimpe
Fast feedback control and safety guarantees are essential in modern robotics.
1 code implementation • L4DC 2020 • Mona Buisson-Fenet, Friedrich Solowjow, Sebastian Trimpe
Despite the availability of ever more data enabled through modern sensor and computer technology, it still remains an open problem to learn dynamical systems in a sample-efficient way.
no code implementations • 29 Oct 2019 • Matteo Turchetta, Andreas Krause, Sebastian Trimpe
In reinforcement learning (RL), an autonomous agent learns to perform complex tasks by maximizing an exogenous reward signal while interacting with its environment.
1 code implementation • 7 Oct 2019 • Steve Heim, Alexander von Rohr, Sebastian Trimpe, Alexander Badri-Spröwitz
While safety can only be guaranteed after learning the safety measure, we show that failures can already be greatly reduced by using the estimated measure during learning.
no code implementations • 24 Jul 2019 • Alonso Marco, Dominik Baumann, Philipp Hennig, Sebastian Trimpe
Learning robot controllers by minimizing a black-box objective cost using Bayesian optimization (BO) can be time-consuming and challenging.
2 code implementations • 14 May 2019 • Andreas Doerr, Michael Volpp, Marc Toussaint, Sebastian Trimpe, Christian Daniel
Policy gradient methods are powerful reinforcement learning algorithms and have been demonstrated to solve many complex tasks.
no code implementations • 15 Dec 2018 • Matthias Neumann-Brosig, Alonso Marco, Dieter Schwarzmann, Sebastian Trimpe
Bayesian optimization is proposed for automatic learning of optimal controller parameters from experimental data.
1 code implementation • 13 Sep 2018 • Dominik Baumann, Jia-Jie Zhu, Georg Martius, Sebastian Trimpe
Event-triggered control (ETC) methods can achieve high-performance control with a significantly lower number of samples compared to usual, time-triggered methods.
no code implementations • 10 Sep 2018 • Alexander von Rohr, Sebastian Trimpe, Alonso Marco, Peer Fischer, Stefano Palagi
Soft microrobots based on photoresponsive materials and controlled by light fields can generate a variety of different gaits.
no code implementations • 11 Jun 2018 • Michael Hertneck, Johannes Köhler, Sebastian Trimpe, Frank Allgöwer
A supervised learning framework is proposed to approximate a model predictive controller (MPC) with reduced computational complexity and guarantees on stability and constraint satisfaction.
no code implementations • 27 May 2018 • Arash Mehrjou, Friedrich Solowjow, Sebastian Trimpe, Bernhard Schölkopf
Apart from its application for encoding a sequence of observations, we propose to use the compression achieved by this encoding as a criterion for model selection.
no code implementations • 5 Mar 2018 • Friedrich Solowjow, Dominik Baumann, Jochen Garcke, Sebastian Trimpe
Common event-triggered state estimation (ETSE) algorithms save communication in networked control systems by predicting agents' behavior, and transmitting updates only when the predictions deviate significantly.
4 code implementations • ICML 2018 • Andreas Doerr, Christian Daniel, Martin Schiegg, Duy Nguyen-Tuong, Stefan Schaal, Marc Toussaint, Sebastian Trimpe
State-space models (SSMs) are a highly expressive model class for learning patterns in time series data and for system identification.
no code implementations • 20 Sep 2017 • Alonso Marco, Philipp Hennig, Stefan Schaal, Sebastian Trimpe
Finding optimal feedback controllers for nonlinear dynamic systems from data is hard.
no code implementations • 8 Mar 2017 • Andreas Doerr, Duy Nguyen-Tuong, Alonso Marco, Stefan Schaal, Sebastian Trimpe
PID control architectures are widely used in industrial applications.
no code implementations • 3 Mar 2017 • Alonso Marco, Felix Berkenkamp, Philipp Hennig, Angela P. Schoellig, Andreas Krause, Stefan Schaal, Sebastian Trimpe
In practice, the parameters of control policies are often tuned manually.
no code implementations • 6 May 2016 • Alonso Marco, Philipp Hennig, Jeannette Bohg, Stefan Schaal, Sebastian Trimpe
With this framework, an initial set of controller gains is automatically improved according to a pre-defined performance objective evaluated from experimental data.
1 code implementation • 19 Feb 2016 • Jan Issac, Manuel Wüthrich, Cristina Garcia Cifuentes, Jeannette Bohg, Sebastian Trimpe, Stefan Schaal
To address this issue, we show how a recently published robustification method for Gaussian filters can be applied to the problem at hand.
no code implementations • 14 Sep 2015 • Manuel Wüthrich, Cristina Garcia Cifuentes, Sebastian Trimpe, Franziska Meier, Jeannette Bohg, Jan Issac, Stefan Schaal
The contribution of this paper is to show that any Gaussian filter can be made compatible with fat-tailed sensor models by applying one simple change: Instead of filtering with the physical measurement, we propose to filter with a pseudo measurement obtained by applying a feature function to the physical measurement.