This paper addresses the problem of learning optimal policies for satisfying signal temporal logic (STL) specifications by agents with unknown stochastic dynamics. The system is modeled as a Markov decision process, in which states represent partitions continuous space and transition probabilities are unknown. We formulate two synthesis problems where desired STL specification enforced maximizing probability satisfaction, expected robustness degree, that is, measure quantifying quality...

The increased complexity of modern systems necessitates automated anomaly detection methods to detect possible anomalous behavior determined by malfunctions or external attacks. We present formal for inferring (via supervised learning) and detecting unsupervised behavior. Our procedures use data construct a signal temporal logic (STL) formula that describes normal system This can be used formulate properties such as "If the train brakes within 500 m platform at speed 50 km/hr, then it will...

Networked dynamical systems are increasingly used as models for a variety of processes ranging from robotic teams to collections genetically engineered living cells. As the complexity these increases, so does range emergent properties that they exhibit. In this work, we define new logic called Spatial-Temporal Logic (SpaTeL) is unification signal temporal (STL) and tree spatial superposition (TSSL). SpaTeL capable describing high-level patterns change over time, e.g., "Power consumption in...

As the complexity of cyber-physical systems increases, so does number ways an adversary can disrupt them. This necessitates automated anomaly detection methods to detect possible threats. In this paper, we extend our recent results in field inference via formal develop unsupervised learning algorithm. Our procedure constructs from data a signal temporal logic (STL) formula that describes normal system behavior. Trajectories do not satisfy learned are flagged as anomalous. STL be used...

This paper presents an inference algorithm that can discover temporal logic properties of a system from data. Our operates on finite time trajectories are labeled according to whether or not they demonstrate some desirable (e.g. "the car successfully stops before hitting obstruction"). A formula discriminate between the behaviors and undesirable ones is constructed. The formulae also indicate possible causes for each set "If speed greater than 15 m/s within 0.5s brake application,...

Abstract Dynamical system models have proven useful for decoding the current brain state from neural activity. So far, neuroscience has largely relied on either linear or nonlinear based artificial networks. Piecewise approximations of dynamics in other technical applications, providing a clear advantage over network-based models, when dynamical is not only supposed to be observed, but also controlled. Here we explore whether piecewise-linear (recurrent Switching Linear System rSLDS models)...

The maturity of sensor network technologies has facilitated the emergence an industrial Internet Things (IIoT), which collected increasing volume data. Converting these data into actionable intelligence for fault diagnosis is key to reducing unscheduled downtime and performance degradation, among other examples. This article formalizes a problem called semantic diagnosis- construct formal specifications faults directly from IIoT-enabled systems. are written as signal temporal logic formulas,...

Electric Vertical-Take-Off-and-Landing aircraft has been emerging as a revolutionary transportation mode. A major limiting factor for unmanned and manned applications is the energy performance, which determines flight time range. Characterization modeling of underlying multi-physical dynamics critical efforts on design, motion planning, control to improve often take model-based approach. system-level model incorporating all relevant subsystem their coupling missing in literature. To fill...

Smoke plumes emitted from wildland-urban interface (WUI) wildfires contain toxic chemical substances that are harmful to human health, mainly due the burning of synthetic components. Accurate measurement these air toxics is necessary for understanding their impacts on health. However, pollution typically measured using ground-based sensors, manned airplanes, or satellites, which all provide low-resolution data. Unmanned Aerial Vehicles (UAVs) have potential high-resolution spatial and...

This paper introduces an innovative approach to 3D environmental mapping through the integration of a compact, handheld sensor package with two-stage fusion pipeline. The package, incorporating LiDAR, IMU, RGB, and thermal cameras, enables comprehensive robust various environments. By leveraging Simultaneous Localization Mapping (SLAM) imaging, our solution offers good performance in conditions where global positioning is unavailable visually degraded runs real-time LiDAR-Inertial SLAM...

This paper presents the foundations for analysis and modeling of human guidance behavior that is based on emergent patterns in closed-loop agent-environment dynamics. The central hypothesis these patterns, which can be explained terms invariants inherent to dynamics, provide building blocks organization behavior. concept interaction first introduced using a toy example then detailed formally dynamical system control principles. demonstrates existence significance data collected experiments...

This paper describes a finite-horizon receding horizon trajectory optimization scheme which uses an approximation of the value function to provide cost-to-go (CTG) and associated state information. The is computed using finite-state, motion primitive automaton vehicle dynamics. Using actual instead heuristic CTG allows tighter integration between planning control layers needed for vehicles operating in challenging spatial environments. It also enables more rigorous use framework autonomous...

Modeling agile and versatile spatial behavior remains a challenging task, due to the intricate coupling of planning, control, perceptual processes. Previous results have shown that humans plan organize their guidance by exploiting patterns in interactions between agent or organism environment. These patterns, described under concept Interaction Patterns (IPs), capture invariants arising from equivalences symmetries interaction with environment, as well effects intrinsic properties human...

Abstract Flying animals accomplish high-speed navigation through fields of obstacles using a suite sensory modalities that blend spatial memory with input from vision, tactile sensing, and, in the case most bats and some other animals, echolocation. Although good deal previous research has been focused on role individual modes sensing animal locomotion, our understanding integration interplay among is still meager. To understand how integrate echolocation, vision memory, we conducted an...

Multiple-player games involving cooperative and adversarial agents are a type of problems great practical significance. In this letter, we consider an attack-defense game with single attacker multiple defenders. The attempts to enter protected region, while the defenders attempt defend same region capture outside region. We propose distributed pursuit-defense strategy for defenders' defense against attacker. Inside bounded, convex, two-dimensional space, choose among area-decreasing,...

This paper describes a mapping method for the analysis of guidance performance. Spatial state and time-to-go maps, along with their statistics, are computed from an ensemble trajectories. The technique is motivated by concept spatial value function associated optimal model. For illustration, applied to trajectories collected human-operated miniature helicopter in precision interception task. closed-loop dynamics under human control was modeled as mass-point system. model provides formal...

This paper uses active learning to solve the problem of mining signal temporal requirements cyber-physical systems or simply requirement problem. By utilizing robustness degree, we formulate as an optimization We then propose a new algorithm called Gaussian Process Adaptive Confidence Bound (GP-ACB) help in solving show theoretically that GP-ACB has lower regret bound-thus larger convergence rate-than some existing algorithms, such GP-UCB. finally illustrate and apply our with two case...