Search

Short-term weather forecasting, i.e., nowcasting, is an important area of research, with application to logistics, insurance, and environmental and social governance. More accurate forecasting is required to achieve nowcasting, which implies the need for improvement in either the acquisition of weather data or the development of better forecasting algorithms. While much of the literature focuses on algorithmic improvements, we consider how one may incentivize better data acquisition. Our idea is to implement a market-based incentive scheme that rewards meteorological data acquisition in proportion to how useful that data is in improving weather forecasts.Our scheme incentivizes optimal placement of meteorological sensors by issuing blockchain-backed digital currency in direct proportion to the sensitivity of the weather forecast to raw data inputs. The scheme consists of an algorithm whose outputs are weather forecasts and whose inputs are data from weather sensors. The algorithm issues digital tokens in direct proportion to the sensitivity of outputs, i.e., the errors in forecasting, to inputs, i.e., sensor data. The issuance has the effects of incentivizing better sensor placement and improved sensor quality. Placement is incentivized because poor placement results in a lower forecast sensitivity; for example, this could happen due to placement near a non-representative microclimate, causing an aberration, or near another, well-placed sensor, resulting in redundancy. Quality is incentivized because higher certainty results in higher forecast sensitivity. With improved sensor placement and quality, the scheme therefore incentivizes better weather forecasts.The incentivization therefore encourages better data assimilation and data availability and also enables the development of solutions that encourage better forecasting with market mechanisms. For example, a digital token may be traded between market participants and tokens may be issued by some authority to incentivize improved data acquisition. This latter example can be extended to a situation where an authority seeks to improve data acquisition in an area with low monitoring frequency, such as the world's oceans.We present the results of a demonstration performed using a network of three Intellisense weather stations with forecasts based on a modification of the NAM‑NMM, performing nowcasts for locations in Torrance, California. The model was implemented in an oracle that performs forecasts and uses the results of the forecasts to issue rewards on the Solana blockchain testnet. The three weather stations were place in two different locations, with the two sensors placed in the same location having different update frequencies, resulting in higher uncertainty in the data coming from the lower-fidelity sensor. The two sensors in the same location cannibalize each other's earnings, and the higher-fidelity sensor is rewarded with the larger share of earnings.

We consider the orbital maintenance problem on a quasi-satellite orbit about the Martian moon, Phobos. The orbit is computed using a high-fidelity ephemeris model so that the major sources of disturbances are due to measurement error. Two types of orbit maintenance schemes are considered. The first is based on asymptotically tracking the desired trajectory and the second is based on stabilizing to the manifold of trajectories that share the same Jacobi constant as the reference trajectory. The latter can be done because trajectories with the same Jacobi constant are in the neighborhood of one another. The results show that the trajectory-tracking scheme has lower fuel consumption when tracking must be precise and that the approach of stabilizing to a manifold has better fuel consumption at the expense of tracking.

In this work, we introduce a scalable, decentralized deep reinforcement learning (DRL) scheme for controlling traffic signalization. The work builds on previous results using multi-agent DRL, with a new state representation and reward definitions. The state representation is a coarse image of traffic and the definitions of reward functions are tested based on the simulated Monaco SUMO Traffic (MoST) scenario. Based on extensive numerical experimentation, we have found the most appropriate choice of the reward function is related to minimizing the average amount of time vehicles spent in the network, but with various modifications that improve the learning process. The resulting algorithm performs better than the previous one on which it is based and markedly better than a non-learning based, greedy policy.

Future spacecraft missions require novel guidance and control policies that reduce fuel consumption, yield sparse thrust signals, and maintain mission safety even in the presence of faults. This paper presents an approach for rendezvous with a target in a near-rectilinear halo orbit that exploits the natural dynamics to reduce propellant consumption, and ensures passive safety in the presence of thruster failure. A chaser spacecraft that aims to rendezvous with a target is steered into coasting sets while simultaneously maintaining passive safety by avoiding the states that naturally collide with the target. Upon entering the coasting sets, the chaser's thrusters are disengaged, as the natural dynamics lead it into a goal set. Abort-safety is then demonstrated during final approach from the goal set to the target. The target is modeled in a full-ephemeris, high-fidelity, and quasi-periodic near-rectilinear halo orbit. Simulations demonstrate a reduction in maneuver fuel consumption, measured by delta-v, of up to 72.5% and a significant reduction of thruster on-time compared to prior work.

We present a scheme for verifiable location reporting of small satellites, or small sats. The decreasing size of small sats makes them harder to track optically. We propose to verify satellite locations by having the satellites verify the correctness of each other’s positions by occasionally completing cryptographically-secure, telemetry-based challenges that are secured on a permissioned blockchain.In the scheme, we update an a posteriori position estimate using self-reporting. The self-reported data is verified by requiring that satellites periodically complete challenges through which it becomes possible to determine the correctness of position using bilateration. To avoid equivocation, the challenges are cryptographically secured by wrapping a challenge in layers that can be removed only by the satellite that is targeted by the challenger. The blockchain is used to secure a list of locations so that verification can be done asynchronously. This is necessary because satellite communication can be significantly delayed. The blockchain is permissioned and operated by constellations, not individual satellites, and does not require that any satellite host a node.

Two-stage processing machines for manufacturing are equipped with two sets of actuators for motion control with different operating ranges and bandwidths. The processing time, product quality, and flexibility of the manufacturing process can be optimized by coordinating these actuators and exploiting the entire actuator operating range. We propose a motion control strategy for two-stage processing machines based on the model predictive control (MPC). By exploiting timescale separation, we formulate the problem as a single-stage motion control with reference-dependent constraints. The feasibility of the MPC problem is guaranteed by using a reference governor that adjusts the feed rate. The proposed method guarantees correct processing while satisfying the actuators’ range and dynamics constraints, finite-time processing of a given spatial pattern, and real-time execution even with limited computational resources. Simulation and experimental results on a real processing pattern are shown for a scaled laboratory demonstration machine.

We provide a scheme for verifying satellite locations, where the locations are reported by satellites themselves. In the scheme, satellites periodically report their locations to the network and these are included in a list. The locations are occasionally verified through a type of proof-of-location protocol. A proof of location is provided via the solution to a cryptographic puzzle and adherence to geometric constraints. The cryptographic puzzle is constructed as a layered data packet, where each layer can be removed only by a specific satellite. This by itself ensures that the challenge was completed in the requisite order. As part of the response to the challenge, each satellite reports its position and time. These reports are verified against the physical constraints of the speed of light and radio signal strength decay.The scheme secures location reports and proofs of location using a permissioned blockchain. Having a source of truth allows locations to be verified retrospectively, which is important due to asynchronous communication between satellites. Satellite reports can arrive out of order and updating past observations with more reliable information is essential to achieving consensus on the veracity of location reports. Although location verification could be implemented as a centralized scheme, decentralization enables scalability and trustless cooperation between satellite operators, potentially reducing costs in deployment and operation of satellite constellation.

We consider the problem of incentivization and optimal control of autonomous vehicles for improving traffic congestion. In our scenario, autonomous vehicles must be incentivized in order to participate in traffic improvement. Using the theory and methods of optimal transport, we propose a constrained optimization framework over dynamics governed by partial differential equations, so that we can optimally select a portion of vehicles to be incentivized and controlled. The goal of the optimization is to obtain a uniform distribution of vehicles over the spatial domain. To achieve this, we consider two types of penalties on vehicle density, one is the $L^2$ cost and the other is a multiscale-norm cost, commonly used in fluid-mixing problems. To solve this non-convex optimization problem, we introduce a novel algorithm, which iterates between solving a convex optimization problem and propagating the flow of uncontrolled vehicles according to the Lighthill-Whitham-Richards model. We perform numerical simulations, which suggest that the optimization of the $L^2$ cost is ineffective while optimization of the multiscale norm is effective. The results also suggest the use of a dedicated lane for this type of control in practice., 6 pages, 10 figures, submitted to the 2020 IEEE Intelligent Vehicles Symposium

In this brief, we consider the constrained optimization problem underpinning model predictive control (MPC). We show that this problem can be decomposed into an unconstrained optimization problem with the same cost function as the original problem and a constrained optimization problem with a modified cost function and dynamics that have been precompensated according to the solution of the unconstrained problem. In the case of linear systems subject to a quadratic cost, the unconstrained problem has the familiar LQR solution and the constrained problem reduces to a minimum-norm projection. This implies that solving linear MPC problems is equivalent to precompensating a system using LQR and applying MPC to penalize only the control input. We propose to call this a constraint-separation principle and discuss the utility of both constraint separation and general decomposition in the design of MPC schemes and the development of numerical solvers for MPC problems., 8 pages, 2 figures, submitted to Automatica

This work considers the control of a spacecraft in indefinite near-rectilinear halo orbit about the Earth-Moon L2. For indefinite station-keeping, it is important to minimize fuel consumption, while allowing for occasional transfer to a new orbit. The control scheme therefore consists of two components: the first component is the tracking of the nominal NRHO and the second component is an orbit correction maneuver between NRHO trajectories. The nominal NRHO is computed using a multiple-shooting technique that takes into account all forces on the spacecraft whose magnitude is larger than the dominant disturbance forces caused by navigational error. The tracking component is a linear-quadratic regulation scheme that rejects disturbances caused by orbit determination error, using a Lyapunov sublevel set that models the state covariance generated using sequential Kalman filter. The orbit correction maneuver is computed to minimize fuel costs.

In this paper, we provide a solution to the HJB equation associated with LQR-type problems for fully-actuated left-invariant control systems on compact connected Lie groups. We obtain the corresponding algebraic/differential Riccati equation and in turn generalize some of the results in Berkane and Tayebi (2015) to a broader class of Lie groups. Closed-loop stability results are also derived.

This paper proposes a model-based deep reinforcement learning (DRL) algorithm for cooperative adaptive cruise control (CACC) of connected vehicles. Differing from most existing CACC works, we consider a platoon consisting of both human-driven and autonomous vehicles. The humandriven vehicles are heterogeneous and connected via vehicleto-vehicle (V2V) communication and the autonomous vehicles are controlled by a cloud-based centralized DRL controller via vehicle-to-cloud (V2C) communication. To overcome the safety and robustness issues of RL, the algorithm informs lowerlevel controllers of desired headway signals instead of directly controlling vehicle accelerations. The lower-level behavior is modeled according to the optimal velocity model (OVM), which determines vehicle acceleration according to a headway input. Numerical experiments show that the model-based DRL algorithm outperforms its model-free version in both safety and stability of CACC. Furthermore, we study the impact of different penetration ratios of autonomous vehicles on the safety, stability, and optimality of the CACC policy.

This work proposes a Model Predictive Control (MPC) policy for simultaneous station keeping and momentum management of a low-thrust nadir-pointing satellite in geostationary orbit around the Earth. The satellite is equipped with six electrically powered thrusters and three axisymmetric reaction wheels, which must be coordinated to control the satellite's orbital position and, concurrently, unload the wheels' stored angular momentum. The MPC policy enforces constraints that maintain the satellite in a tight latitude and longitude window and in a tight nadir-pointing attitude configuration, while minimizing the delta-v provided by the thrusters. The MPC policy exploits a prediction model of the environmental disturbance forces in order to significantly reduce the delta-v required for station keeping, and enforces constraints determined by the thruster configuration to select control forces and torques that can be generated by the propulsion system. Numerical simulations of the control policy in closed-loop with the satellite nonlinear dynamics under high-precision orbit propagation provided by Systems Tool Kit (STK) that validate the performance of the proposed design in terms of thruster usage and constraint enforcement are presented.

Jump–diffusion processes (JDPs) involve a combination of jumps (Poisson process) and diffusions (Wiener process). JDPs can be used to model large classes of disturbances in engineering applications, such as road disturbances to a car, wind disturbances to an airplane, and system parameter perturbations. This paper develops a road anomaly detector by exploiting an optimal state estimator for systems driven by JDP in combination with the multi-input observer. State estimation with the JDP-based estimator is shown to have better performance than a Kalman filter when jumps, such as potholes and bumps, are present. The road anomaly detector is implemented in an experimental test vehicle and its experimental validation results are reported.

This work considers the problem of minimizing the path-curvature of a unicycle moving between two configurations in SE(2) with free final time. The problem is posed as an optimal control problem and the necessary conditions are derived using the Pontryagin Maximum Principle and Lie-Poisson reduction. Solutions are categorized into three cases corresponding to the value of the Casimir. A numerical solver for obtaining the optimal control is described. Experimental results are reported on a ground robot.

This paper considers the application of a reference governor scheme to a lighter-than-air wind-energy system subject to wind turbulence. The turbulence is treated as the output of the von Karman model, which can be represented by a linear filter with a normally distributed disturbance input. The conventional reference governor, which enforces constraints robustly, has to be heuristically tuned in order to be applied to stochastic systems. In this work, the reference governor is generalized to the stochastic setting in order to handle stochastic disturbances. A simulation is presented, which shows satisfactory constraint enforcement with properties that are superior to the heuristically-tuned reference governor results of the authors’ previous work.

This paper develops a nonlinear model predictive controller for constrained attitude maneuvering of a fully actuated spacecraft with reaction wheels. In the proposed control algorithm, a Lie group ...

In this work, we introduce a scalable, decentralized deep reinforcement learning (RL) scheme for optimizing vehicle traffic consisting of both autonomous and human-driven vehicles. The control inputs to the system are the following distance and lane placement of the autonomous vehicles and the human-vehicles are uncontrolled. One point of novelty of the scheme is that it is trained on images of traffic, where pixels are colored based on how much of them is occupied by human-driven or autonomous vehicles. Another point of novelty is in how the scheme achieves scalable decentralization; it does so by training multiple RL agents, each responsible for its own region of control, but able to at least partially observe neighboring agents’ regions of control. In this way, the scheme is infinitely scalable because an RL agent does not need to communicate with its neighbors, and can be applied in conjunction with systems in which neighboring controllers are not RL-based.We perform a case study of two simulations on a two-lane oval highway in the Simulation of Urban MObility (SUMO) environment. In the first simulation, a single RL agent is applied so that its region of control coincides with an area of traffic congestion. In the second simulation, we apply multiple RL agents covering the entire network. The results of the first simulation show that the single RL agent is able to improve traffic congestion. The results of the second simulation show that the decentralized RL scheme is able to achieve even better results.

Grover's algorithm is one of the most famous algorithms which explicitly demonstrates how the quantum nature can be utilized to accelerate the searching process. In this work, Grover's quantum search problem is mapped to a time-optimal control problem. Resorting to Pontryagin's Minimum Principle we find that the time-optimal solution has the bang-singular-bang structure. This structure can be derived naturally, without integrating the differential equations, using the geometric control technique where Hamiltonians in the Schr\"odinger's equation are represented as vector fields. In view of optimal control, Grover's algorithm uses the bang-bang protocol to approximate the optimal protocol with a minimized number of bang-to-bang switchings to reduce the query complexity. Our work provides a concrete example how Pontryagin's Minimum Principle is connected to quantum computation, and offers insight into how a quantum algorithm can be designed., Comment: 4 figures

In turbocharged gasoline engines, the violation of the compressor surge constraint can be prevented by exploiting predictive control techniques based on the reference governor (RG). In this paper, two RG designs are considered: an inner-and an outer-loop RG. The inner-loop design governs the throttle command, and the outer-loop governs the requested air charge command. Numerical simulations and the experimental vehicle results are presented, showing that both designs enforce the surge constraint without the use of a compressor bypass valve, but that the inner-loop design requires scheduling in order to do so.

The handling of constraints in systems subject to variable time-delay is a challenging problem. It is particularly relevant to Network Control Systems (NCSs) in which a control system is remotely located with respect to the plant to be controlled. In this paper, we develop reference governors for controlling constrained systems subject to variable delays with a particular focus on the application to NCSs. In the proposed approach, which neither exploits nor depends on any explicit synchronization between the plant and the governor, the closed-loop dynamics are modeled by a sampled data system, for which input delays result in additive disturbances with magnitude proportional to the input rate of change. We first develop a network reference governor (netRG) that guarantees constraint enforcement and finite-time convergence for variable time-delay shorter than the sampling period. Then, we extend the network reference governor to the case of output feedback. Finally, we consider the case of long and potentially unbounded delay. The netRG is evaluated in a case study of orientation control of a spacecraft with a flexible appendage.

This paper presents a reference governor formulation that is applicable to systems with stochastic disturbances and achieves constraint satisfaction properties that are analogous to those of conventional reference governors. In particular, the reference governor proposed herein is shown to enforce chance constraints, guarantee a form of eventual recursive feasibility, and guarantee almost-sure convergence to constant, constraint-admissible reference inputs. It can also be applied to enforce constraints for closed-loop systems with state observers. This stands in contrast with traditional reference governor techniques, which must be heuristically tuned in order to achieve a balance between constraint satisfaction and the size of achievable steady-state references in the presence of stochastic disturbances. A numerical example is reported which illustrates the operation of the reference governor for chance-constrained systems and is compared to the conventional, robust approach.

This paper presents a novel reference-governor-based approach to enforcing constraints in open-loop systems, where the constraint enforcement scheme modifies a control signal from a feedback controller. The reference governor is not typically applicable to these systems, since reference governors modify reference signals, which are inputs to a feedback controller.The design is based on duplicating the feedback controller of the closed-loop system and implementing a reference governor to modify the reference input to the duplicate controller. Effectively, this bypasses the nominal controller, whose internal dynamics may not be asymptotically stable. We present a method to ensure stability of the closed-loop system by ceding some control authority to the nominal controller.A numerical example is considered in which the nominal controller is a human operator of a steer-by-wire system. The operator is modeled as a PID controller with unknown parameters. Results show that the inner-loop governor design is able to satisfactorily enforce constraints in such an application.

This paper considers the design of a controller and a constraint-enforcement scheme for application to dualmode, autonomous and manual steering systems. A tracking controller is designed to track a desired pinion angle during autonomous operation, and to provide assistive torque during manual operation. The tracking controller is designed using H ∞ synthesis with tracking made possible via the solution to a full-information output regulation problem. A reference governor scheme is implemented in order to enforce constraints. Numerical simulations are presented corresponding to an aggressive step-steer maneuver in autonomous mode and show strict constraint enforcement.

For control architectures of autonomous and semi-autonomous driving features, we design a vehicle steering controller with limited preview ensuring that the vehicle constraints are satisfied, and that any piecewise clothoidal trajectory, that is possibly generated by a path planner or supervisory algorithm and satisfies constraints on the desired yaw rate and the change of desired yaw rate, is tracked within a preassigned lateral error bound. The design is based on computing a non-maximal, yet polyhedral, robust control invariant (RCI) set for a system subject to bounded disturbances with state-dependent bounds, which also allows to determine the constraints describing the reference trajectories that can be followed. The RCI set is then enforced by model predictive control, where the cost function enforces additional objectives of the vehicle motion.

This paper develops a nonlinear model predictive controller for constrained attitude maneuvering of spacecraft actuated by a reaction wheel assembly. In the proposed control algorithm, a Lie group variational integrator is implemented for the prediction of discrete-time dynamics. The model predictive control problem is defined on SO(3) and converted to an optimal control problem over the prediction horizon. The numerical solver for this optimal control problem is implemented. The developed solver utilizes the Indirect Single Shooting Method to find a solution to the necessary conditions for optimality. The control constraints and exclusion zone constraints are handled using a penalty function approach. Simulation results on a spacecraft model are presented.