the proceedings contain 114 papers. the topics discussed include: interval reachability of nonlinear dynamical systems with neural network controllers;physics-informed model-based reinforcement learning;learning-to-le...
the proceedings contain 114 papers. the topics discussed include: interval reachability of nonlinear dynamical systems with neural network controllers;physics-informed model-based reinforcement learning;learning-to-learn to guide random search: derivative-free meta Blackbox optimization on manifold;guaranteed conformance of neurosymbolic models to natural constraints;guaranteed conformance of neurosymbolic models to natural constraints;safe and efficient reinforcement learning using disturbance-observer-based control barrier functions;learningthe dynamics of autonomous nonlinear delay systems;improving gradient computation for differentiable physics simulation with contacts;policy gradient play with networked agents in Markov potential games;sample complexity bound for evaluating the robust observer’s performance under coprime factors uncertainty;and end-to-end learning to warm-start for real-time quadratic optimization.
Relying on recent research results on Neural ODEs, this paper presents a methodology for the design of state observers for nonlinear systems based on Neural ODEs, learning Luenberger-like observers and their nonlinear...
详细信息
Relying on recent research results on Neural ODEs, this paper presents a methodology for the design of state observers for nonlinear systems based on Neural ODEs, learning Luenberger-like observers and their nonlinear extension (Kazantzis-Kravaris-Luenberger (KKL) observers) for systems with partially-known nonlinear dynamics and fully unknown nonlinear dynamics, respectively. In particular, for tuneable KKL observers, the relationship between the design of the observer and its trade-off between convergence speed and robustness is analysed and used as a basis for improving the robustness of the learning-based observer in training. We illustrate the advantages of this approach in numerical simulations.
Event-triggered communication and control provide high control performance in networked control systems without overloading the communication network. However, most approaches require precise mathematical models of th...
详细信息
Event-triggered communication and control provide high control performance in networked control systems without overloading the communication network. However, most approaches require precise mathematical models of the system dynamics, which may not always be available. Model-free learning of communication and control policies provides an alternative. Nevertheless, existing methods typically consider single-agent settings. this paper proposes a model-free reinforcement learning algorithm that jointly learns resource-aware communication and control policies for distributed multi-agent systems from data. We evaluate the algorithm in a high-dimensional and nonlinear simulation example and discuss promising avenues for further research.
fWe introduce a method for controlling systems with nonlinear dynamics and full actuation by approximating the dynamics with polynomials and applying a system level synthesis controller. We show how to optimize over t...
详细信息
fWe introduce a method for controlling systems with nonlinear dynamics and full actuation by approximating the dynamics with polynomials and applying a system level synthesis controller. We show how to optimize over this class of controllers using a neural network while maintaining stability guarantees, without requiring a Lyapunov function. We give bounds for the domain over which the use of the class of controllers preserves stability and gives bounds on the control costs incurred by optimized controllers. We then numerically validate our approach and show improved performance compared with feedback linearization- suggesting that the SLS controllers are able to take advantage of nonlinearities in the dynamics while guaranteeing stability.
Transfer operators offer linear representations and global, physically meaningful features of nonlinear dynamical systems. Discovering transfer operators, such as the Koopman operator, require careful crafted dictiona...
详细信息
Transfer operators offer linear representations and global, physically meaningful features of nonlinear dynamical systems. Discovering transfer operators, such as the Koopman operator, require careful crafted dictionaries of observables, acting on states of the dynamical system. this is ad hoc and requires the full dataset for evaluation. In this paper, we offer an optimization scheme to allow joint learning of the observables and Koopman operator with online data. Our results show we are able to reconstruct the evolution and represent the global features of complex dynamical systems.
Safety is one of the biggest concerns to applying reinforcement learning (RL) to the physical world. In its core part, it is challenging to ensure RL agents persistently satisfy a hard state constraint without white-b...
详细信息
Safety is one of the biggest concerns to applying reinforcement learning (RL) to the physical world. In its core part, it is challenging to ensure RL agents persistently satisfy a hard state constraint without white-box or black-box dynamics models. this paper presents an integrated model learning and safe control framework to safeguard any RL agent, where the environment dynamics are learned as Gaussian processes. the proposed theory provides (i) a novel method to construct an offline dataset for model learningthat best achieves safety requirements;(ii) a design rule to construct the safety index to ensure the existence of safe control under control limits;(iii) a probablistic safety guarantee (i.e. probabilistic forward invariance) when the model is learned using the aforementioned dataset. Simulation results show that our framework achieves almost zero safety violation on various continuous control tasks.
We consider the problem of online control of systems with time-varying linear dynamics. To state meaningful guarantees over changing environments, we introduce the metric of adaptive regret to the field of control. th...
详细信息
We consider the problem of online control of systems with time-varying linear dynamics. To state meaningful guarantees over changing environments, we introduce the metric of adaptive regret to the field of control. this metric, originally studied in online learning, measures performance in terms of regret against the best policy in hindsight on any interval in time, and thus captures the adaptation of the controller to changing dynamics. Our main contribution is a novel efficient meta-algorithm: it converts a controller with sublinear regret bounds into one with sublinear adaptive regret bounds in the setting of time-varying linear dynamical systems. the underlying technical innovation is the first adaptive regret bound for the more general framework of online convex optimization with memory. Furthermore, we give a lower bound showing that our attained adaptive regret bound is nearly tight for this general framework.
this paper proposes a Bellman Deviation algorithm for the detection of man-in-the-middle (MITM) attacks occurring when an agent controls a Markov Decision Process (MDP) system using model-free reinforcement learning. ...
详细信息
this paper proposes a Bellman Deviation algorithm for the detection of man-in-the-middle (MITM) attacks occurring when an agent controls a Markov Decision Process (MDP) system using model-free reinforcement learning. this algorithm is derived by constructing a "Bellman Deviation sequence" and findind stochastic bounds on its running sequence average. We show that an intuitive, necessary and sufficient "informational advantage" condition must be met for the proposed algorithm to guarantee the detection of attacks with high probability, while also avoiding false alarms.
this paper considers online switching control with a finite candidate controller pool, an unknown dynamical system, and unknown cost functions. the candidate controllers can be unstabilizing policies. We only require ...
详细信息
this paper considers online switching control with a finite candidate controller pool, an unknown dynamical system, and unknown cost functions. the candidate controllers can be unstabilizing policies. We only require at least one candidate controller to satisfy certain stability properties, but we do not know which one is stabilizing. We design an online algorithm that guarantees finite-gain stability throughout the duration of its execution. We also provide a sublinear policy regret guarantee compared withthe optimal stabilizing candidate controller. Lastly, we numerically test our algorithm on quadrotor planar flights and compare it with a classical switching control algorithm, falsification-based switching, and a classical multi-armed bandit algorithm, Exp3 with batches.
We present DLKoopman - a software package for Koopman theory that uses deep learning to learn an encoding of a nonlinear dynamical system into a linear space, while simultaneously learningthe linear dynamics. While s...
详细信息
We present DLKoopman - a software package for Koopman theory that uses deep learning to learn an encoding of a nonlinear dynamical system into a linear space, while simultaneously learningthe linear dynamics. While several previous efforts have either restricted the ability to learn encodings, or been bespoke efforts designed for specific systems, DLKoopman is a generalized tool that can be applied to data-driven learning and analysis of any dynamical system. It can either be trained on data from individual states (snapshots) of a system and used to predict its unknown states, or trained on data from trajectories of a system and used to predict unknown trajectories for new initial states. DLKoopman is available on the Python Package Index (PyPI) as 'dlkoopman', and includes extensive documentation and tutorials. Additional contributions of the package include a novel metric called Average Normalized Absolute Error for evaluating performance, and a ready-to-use hyperparameter search module for improving performance.
暂无评论