Search

Learning how to effectively control unknown dynamical systems is crucial for intelligent autonomous systems. This task becomes a significant challenge when the underlying dynamics are changing with time. Motivated by this challenge, this paper considers the problem of controlling an unknown Markov jump linear system (MJS) to optimize a quadratic objective. By taking a model-based perspective, we consider identification-based adaptive control for MJSs. We first provide a system identification algorithm for MJS to learn the dynamics in each mode as well as the Markov transition matrix, underlying the evolution of the mode switches, from a single trajectory of the system states, inputs, and modes. Through mixing-time arguments, sample complexity of this algorithm is shown to be $\mathcal{O}(1/\sqrt{T})$. We then propose an adaptive control scheme that performs system identification together with certainty equivalent control to adapt the controllers in an episodic fashion. Combining our sample complexity results with recent perturbation results for certainty equivalent control, we prove that when the episode lengths are appropriately chosen, the proposed adaptive control scheme achieves $\mathcal{O}(\sqrt{T})$ regret, which can be improved to $\mathcal{O}(polylog(T))$ with partial knowledge of the system. Our proof strategy introduces innovations to handle Markovian jumps and a weaker notion of stability common in MJSs. Our analysis provides insights into system theoretic quantities that affect learning accuracy and control performance. Numerical simulations are presented to further reinforce these insights.

Starting from an integral of the interaction region of colliding Einstein-Maxwell waves and by applying a coordinate transformation, we obtain the charged version of the static Zipoy-Voorhees (ZV) metric valid for all values of the distortion parameter $\gamma$. In Schwarzschild coordinates, we investigate the effect of the charge in the newly found spacetime, stress the analogy with Reissner - Nordstrom metric and discuss some of its features. It is shown that from the expression of Weyl curvature, directional singularities become manifest. For astrophysical importance, we find lensing of null geodesics from the Gauss - Bonnet theorem in such non - spherically charged objects. To prepare the ground for our null, circular geodesics we consider the angular equation linearized about the symmetry plane $\theta=\pi/2$. This, in turn, suggests the distortion parameter (the $ZV$ parameter) must be in the interval $1/2<\gamma<1$. It is found that the lensing angle is highly dependent on the distortion parameter, and becomes decisive on the effect of the charge. For a class of charged compact stars, we plot the deflection angle versus the ratio of impact parameter to the radius of the star. Plots have revealed that for perfectly spherical compact stars, it is hard to identify the effect of electric/magnetic charge, but for non-spherical compact stars the effect of electric charge becomes apparent. For comparison, the same lensing angle has also been found for the stationary ZV metric in the equatorial plane. Our analysis indicates that depending on the value of $\gamma$, the stationary state induces counter effect on the bending angle and thus, when compared with the uncharged static ZV case, the bending angle decreases. The influence of the parameter $\gamma$ on the gravitational redshift is also displayed., Comment: New sections are added and supported with new graphs. Typos are corrected. Matched with published version

In this paper, we derive closed-form expressions for implicit controlled invariant sets for discrete-time controllable linear systems with measurable disturbances. In particular, a disturbance-reactive (or disturbance feedback) controller in the form of a parameterized finite automaton is considered. We show that, for a class of automata, the robust positively invariant sets of the corresponding closed-loop systems can be expressed by a set of linear inequality constraints in the joint space of system states and controller parameters. This leads to an implicit representation of the invariant set in a lifted space. We further show how the same parameterization can be used to compute invariant sets when the disturbance is not available for measurement., Comment: 10 pages, accepted by CDC 2021

Zero-shot domain adaptation (ZDA) methods aim to transfer knowledge about a task learned in a source domain to a target domain, while data from target domain are not available. In this work, we address learning feature representations which are invariant to and shared among different domains considering task characteristics for ZDA. To this end, we propose a method for task-guided ZDA (TG-ZDA) which employs multi-branch deep neural networks to learn feature representations exploiting their domain invariance and shareability properties. The proposed TG-ZDA models can be trained end-to-end without requiring synthetic tasks and data generated from estimated representations of target domains. The proposed TG-ZDA has been examined using benchmark ZDA tasks on image classification datasets. Experimental results show that our proposed TG-ZDA outperforms state-of-the-art ZDA methods for different domains and tasks.

In this paper, we study feedback dynamical systems with memoryless controllers under imperfect information. We develop an algorithm that searches for "adversarial scenarios", which can be thought of as the strategy for the adversary representing the noise and disturbances, that lead to safety violations. The main challenge is to analyze the closed-loop system's vulnerabilities with a potentially complex or even unknown controller in the loop. As opposed to commonly adopted approaches that treat the system under test as a black-box, we propose a synthesis-guided approach, which leverages the knowledge of a plant model at hand. This hence leads to a way to deal with gray-box systems (i.e., with known plant and unknown controller). Our approach reveals the role of the imperfect information in the violation. Examples show that our approach can find non-trivial scenarios that are difficult to expose by random simulations. This approach is further extended to incorporate model mismatch and to falsify vision-in-the-loop systems against finite-time reach-avoid specifications., Comment: This article appears as part of the ESWEEK-TECS special issue and was presented in the International Conference on Embedded Software (EMSOFT)

We revisit the problem of computing (robust) controlled invariant sets for discrete-time linear systems. Departing from previous approaches, we consider implicit, rather than explicit, representations for controlled invariant sets. Moreover, by considering such representations in the space of states and finite input sequences we obtain closed-form expressions for controlled invariant sets. An immediate advantage is the ability to handle high-dimensional systems since the closed-form expression is computed in a single step rather than iteratively. To validate the proposed method, we present thorough case studies illustrating that in safety-critical scenarios the implicit representation suffices in place of the explicit invariant set. The proposed method is complete in the absence of disturbances, and we provide a weak completeness result when disturbances are present., Comment: 28 pages, 9 figures, 3 tables

Zonotopes are widely used for over-approximating forward reachable sets of uncertain linear systems for verification purposes. In this paper, we use zonotopes to achieve more scalable algorithms that under-approximate backward reachable sets of uncertain linear systems for control design. The main difference is that the backward reachability analysis is a two-player game and involves Minkowski difference operations, but zonotopes are not closed under such operations. We under-approximate this Minkowski difference with a zonotope, which can be obtained by solving a linear optimization problem. We further develop an efficient zonotope order reduction technique to bound the complexity of the obtained zonotopic under-approximations. The proposed approach is evaluated against existing approaches using randomly generated instances and illustrated with several examples.

Real-world control applications often involve complex dynamics subject to abrupt changes or variations. Markov jump linear systems (MJS) provide a rich framework for modeling such dynamics. Despite an extensive history, theoretical understanding of parameter sensitivities of MJS control is somewhat lacking. Motivated by this, we investigate robustness aspects of certainty equivalent model-based optimal control for MJS with quadratic cost function. Given the uncertainty in the system matrices and in the Markov transition matrix is bounded by $\epsilon$ and $\eta$ respectively, robustness results are established for (i) the solution to coupled Riccati equations and (ii) the optimal cost, by providing explicit perturbation bounds which decay as $\mathcal{O}(\epsilon + \eta)$ and $\mathcal{O}((\epsilon + \eta)^2)$ respectively., Comment: 17 pages, 8 figures

Federated Learning (FL) is a framework which enables distributed model training using a large corpus of decentralized training data. Existing methods aggregate models disregarding their internal representations, which are crucial for training models in vision tasks. System and statistical heterogeneity (e.g., highly imbalanced and non-i.i.d. data) further harm model training. To this end, we introduce a method, called FedProto, which computes client deviations using margins of prototypical representations learned on distributed data, and applies them to drive federated optimization via an attention mechanism. In addition, we propose three methods to analyse statistical properties of feature representations learned in FL, in order to elucidate the relationship between accuracy, margins and feature discrepancy of FL models. In experimental analyses, FedProto demonstrates state-of-the-art accuracy and convergence rate across image classification and semantic segmentation benchmarks by enabling maximum margin training of FL models. Moreover, FedProto reduces uncertainty of predictions of FL models compared to the baseline. To our knowledge, this is the first work evaluating FL models in dense prediction tasks, such as semantic segmentation., Comment: 11 pages manuscript, 6 pages supplemental material

We present a method for contraction-based feedback motion planning of locally incrementally exponentially stabilizable systems with unknown dynamics that provides probabilistic safety and reachability guarantees. Given a dynamics dataset, our method learns a deep control-affine approximation of the dynamics. To find a trusted domain where this model can be used for planning, we obtain an estimate of the Lipschitz constant of the model error, which is valid with a given probability, in a region around the training data, providing a local, spatially-varying model error bound. We derive a trajectory tracking error bound for a contraction-based controller that is subjected to this model error, and then learn a controller that optimizes this tracking bound. With a given probability, we verify the correctness of the controller and tracking error bound in the trusted domain. We then use the trajectory error bound together with the trusted domain to guide a sampling-based planner to return trajectories that can be robustly tracked in execution. We show results on a 4D car, a 6D quadrotor, and a 22D deformable object manipulation task, showing our method plans safely with learned models of high-dimensional underactuated systems, while baselines that plan without considering the tracking error bound or the trusted domain can fail to stabilize the system and become unsafe., Comment: Extended paper; abridged version presented at the 60th IEEE Conference on Decision and Control (CDC 2021)

Two established approaches to engineer adaptive systems are architecture-based adaptation that uses a Monitor-Analysis-Planning-Executing (MAPE) loop that reasons over architectural models (aka Knowledge) to make adaptation decisions, and control-based adaptation that relies on principles of control theory (CT) to realize adaptation. Recently, we also observe a rapidly growing interest in applying machine learning (ML) to support different adaptation mechanisms. While MAPE and CT have particular characteristics and strengths to be applied independently, in this paper, we are concerned with the question of how these approaches are related with one another and whether combining them and supporting them with ML can produce better adaptive systems. We motivate the combined use of different adaptation approaches using a scenario of a cloud-based enterprise system and illustrate the analysis when combining the different approaches. To conclude, we offer a set of open questions for further research in this interesting area., Comment: 7 pages

Incorporating predictions of external inputs, which can otherwise be treated as disturbances, has been widely studied in control and computer science communities. These predictions are commonly referred to as preview in optimal control and lookahead in temporal logic synthesis. However, little work has been done for analyzing the value of preview information for safety control for systems with continuous state spaces. In this work, we start from showing general properties for discrete-time nonlinear systems with preview and strategies on how to determine a good preview time, and then we study a special class of linear systems, called systems in Brunovsky canonical form, and show special properties for this class of systems. In the end, we provide two numerical examples to further illustrate the value of preview in safety control., Comment: 10 pages, accepted by ACC 2021

Opacity is an information flow property that captures the notion of plausible deniability in dynamic systems, that is whether an intruder can deduce that "secret" behavior has occurred. In this paper we provide a general framework of opacity to unify the many existing notions of opacity that exist for discrete event systems. We use this framework to discuss language-based and state-based notions of opacity over automata. We present several methods for language-based opacity verification, and a general approach to transform state-based notions into language-based ones. We demonstrate this approach for current-state and initial-state opacity, unifying existing results. We then investigate the notions of K-step opacity. We provide a language-based view of K-step opacity encompassing two existing notions and two new ones. We then analyze the corresponding language-based verification methods both formally and with numerical examples. In each case, the proposed methods offer significant reductions in runtime and space complexity.

We propose a computationally efficient $G$-invariant neural network that approximates functions invariant to the action of a given permutation subgroup $G \leq S_n$ of the symmetric group on input data. The key element of the proposed network architecture is a new $G$-invariant transformation module, which produces a $G$-invariant latent representation of the input data. Theoretical considerations are supported by numerical experiments, which demonstrate the effectiveness and strong generalization properties of the proposed method in comparison to other $G$-invariant neural networks., Comment: Presented as contributed talk at NeurIPS 2020 workshop on Differential Geometry meets Deep Learning

Being able to rapidly respond to the changing scenes and traffic situations by generating feasible local paths is of pivotal importance for car autonomy. We propose to train a deep neural network (DNN) to plan feasible and nearly-optimal paths for kinematically constrained vehicles in small constant time. Our DNN model is trained using a novel weakly supervised approach and a gradient-based policy search. On real and simulated scenes and a large set of local planning problems, we demonstrate that our approach outperforms the existing planners with respect to the number of successfully completed tasks. While the path generation time is about 40 ms, the generated paths are smooth and comparable to those obtained from conventional path planners.

We present a method for learning to satisfy uncertain constraints from demonstrations. Our method uses robust optimization to obtain a belief over the potentially infinite set of possible constraints consistent with the demonstrations, and then uses this belief to plan trajectories that trade off performance with satisfying the possible constraints. We use these trajectories in a closed-loop policy that executes and replans using belief updates, which incorporate data gathered during execution. We derive guarantees on the accuracy of our constraint belief and probabilistic guarantees on plan safety. We present results on a 7-DOF arm and 12D quadrotor, showing our method can learn to satisfy high-dimensional (up to 30D) uncertain constraints, and outperforms baselines in safety and efficiency., Comment: 4th Conference on Robot Learning (CoRL 2020)

We present a method for feedback motion planning of systems with unknown dynamics which provides probabilistic guarantees on safety, reachability, and goal stability. To find a domain in which a learned control-affine approximation of the true dynamics can be trusted, we estimate the Lipschitz constant of the difference between the true and learned dynamics, and ensure the estimate is valid with a given probability. Provided the system has at least as many controls as states, we also derive existence conditions for a one-step feedback law which can keep the real system within a small bound of a nominal trajectory planned with the learned dynamics. Our method imposes the feedback law existence as a constraint in a sampling-based planner, which returns a feedback policy around a nominal plan ensuring that, if the Lipschitz constant estimate is valid, the true system is safe during plan execution, reaches the goal, and is ultimately invariant in a small set about the goal. We demonstrate our approach by planning using learned models of a 6D quadrotor and a 7DOF Kuka arm. We show that a baseline which plans using the same learned dynamics without considering the error bound or the existence of the feedback law can fail to stabilize around the plan and become unsafe., Comment: Accepted at RA-L and ICRA 2021. Craig Knuth and Glen Chou contributed equally to this work

We present a method for learning multi-stage tasks from demonstrations by learning the logical structure and atomic propositions of a consistent linear temporal logic (LTL) formula. The learner is given successful but potentially suboptimal demonstrations, where the demonstrator is optimizing a cost function while satisfying the LTL formula, and the cost function is uncertain to the learner. Our algorithm uses the Karush-Kuhn-Tucker (KKT) optimality conditions of the demonstrations together with a counterexample-guided falsification strategy to learn the atomic proposition parameters and logical structure of the LTL formula, respectively. We provide theoretical guarantees on the conservativeness of the recovered atomic proposition sets, as well as completeness in the search for finding an LTL formula consistent with the demonstrations. We evaluate our method on high-dimensional nonlinear systems by learning LTL formulas explaining multi-stage tasks on 7-DOF arm and quadrotor systems and show that it outperforms competing methods for learning LTL formulas from positive examples., Comment: Extended version; conference version to appear in Robotics: Science and Systems XVI (RSS 2020)

Many methods in learning from demonstration assume that the demonstrator has knowledge of the full environment. However, in many scenarios, a demonstrator only sees part of the environment and they continuously replan as they gather information. To plan new paths or to reconstruct the environment, we must consider the visibility constraints and replanning process of the demonstrator, which, to our knowledge, has not been done in previous work. We consider the problem of inferring obstacle configurations in a 2D environment from demonstrated paths for a point robot that is capable of seeing in any direction but not through obstacles. Given a set of \textit{survey points}, which describe where the demonstrator obtains new information, and a candidate path, we construct a Constraint Satisfaction Problem (CSP) on a cell decomposition of the environment. We parameterize a set of obstacles corresponding to an assignment from the CSP and sample from the set to find valid environments. We show that there is a probabilistically-complete, yet not entirely tractable, algorithm that can guarantee novel paths in the space are unsafe or possibly safe. We also present an incomplete, but empirically-successful, heuristic-guided algorithm that we apply in our experiments to 1) planning novel paths and 2) recovering a probabilistic representation of the environment., Comment: Accepted at WAFR 2020

In this paper, we first propose a method that can efficiently compute the maximal robust controlled invariant set for discrete-time linear systems with pure delay in input. The key to this method is to construct an auxiliary linear system (without delay) with the same state-space dimension of the original system in consideration and to relate the maximal invariant set of the auxiliary system to that of the original system. When the system is subject to disturbances, guaranteeing safety is harder for systems with input delays. Ability to incorporate any additional information about the disturbance becomes more critical in these cases. Motivated by this observation, in the second part of the paper, we generalize the proposed method to take into account additional preview information on the disturbances, while maintaining computational efficiency. Compared with the naive approach of constructing a higher dimensional system by appending the state-space with the delayed inputs and previewed disturbances, the proposed approach is demonstrated to scale much better with the increasing delay time., Comment: 7 pages, 3 figures, accepted by ACC 2020