查看更多>>摘要：We introduce a class of partially observed Markov decision processes (POMDPs) with costs that can depend on both the value and (future) uncertainty associated with the initial state. These Initial-State Cost POMDPs (ISC-POMDPs) enable the specification of objectives relative to a priori unknown initial states, which is useful in applications such as robot navigation, controlled sensing, and active perception, that can involve controlling systems to revisit, remain near, or actively infer their initial states. By developing a recursive Bayesian fixed-point smoother to estimate the initial state that resembles the standard recursive Bayesian filter, we show that ISC-POMDPs can be treated as POMDPs with (potentially) belief-dependent costs. We demonstrate the utility of ISC-POMDPs, including their ability to select controls that resolve (future) uncertainty about (past) initial states, in simulation.
查看更多>>摘要：Energy-efficient driving is a key advancement in the deployment of automated vehicles once safety concerns are addressed. This letter formulates the energy-efficient driving problem with constraints and explores various solution methods for common driving scenarios. The findings, rooted in theory of optimal control and Pontryagin’s Minimum Principle (PMP), offer fundamental insights into energy-efficient driving strategies in every-day driving scenarios. Analytical insights from PMP, coupled with fast analytical solution of the respective boundary value problem, enabled implementation in a real-time control system and near-optimal energy savings. The proposed approach was validated through real vehicle testing on the track, with results demonstrating that automated eco-driving can achieve significant energy savings over human drivers in basic daily driving scenarios. This letter not only highlights the effectiveness of the proposed approach but also provides practical guidance for integrating energy-efficient driving strategies into real-world automated driving and advanced driver assistance systems.
查看更多>>摘要：This letter proposes an adaptive event-triggered communication framework for distributed state estimation in sensor networks, enabling each node to self-adapt its transmission rule while maintaining a desired average rate and complying with an upper bound on individual transmission rates. Unlike existing approaches that require manual tuning, the proposed method dynamically tunes transmission thresholds, achieving uniform energy and bandwidth consumption across nodes. We analyze the theoretical properties of the proposed transmission strategy, and verify its effectiveness through simulations in a target-tracking scenario.
查看更多>>摘要：Q-learning and double Q-learning are well-known sample-based, off-policy reinforcement learning algorithms. However, Q-learning suffers from overestimation bias, while double Q-learning suffers from underestimation bias. To address these issues, this letter proposes a weighted smooth Q-learning (WSQL) algorithm. The proposed algorithm employs a weighted combination of the mellowmax operator and the log-sum-exp operator in place of the maximum operator. Firstly, a new stochastic approximation based result is derived and as a consequence the almost sure convergence of the proposed WSQL is presented. Further, a sufficient condition for the boundedness of WSQL algorithm is obtained. Numerical experiments are conducted on benchmark examples to validate the effectiveness of the proposed weighted smooth Q-learning algorithm.
Marta FochesatoFrédy PokouHélène Le CadreJohn Lygeros...
27-32页
查看更多>>摘要：We study noncooperative games with uncertain payoffs, where agents display irrational behaviors in response to underlying risk factors. Our formulation incorporates prospect theory, a behavioral model used to describe agents’ risk attitude, into the equilibrium theory of noncooperative $N-$ agent games. We show that the resulting nonconvex nonsmooth game admits equilibria and provide convergence guarantees for their computation. Further, the concept of “Price of Irrationality” is introduced to quantify the suboptimality induced by irrational behaviors. Finally, we provide bounds on the performance of a class of prospect theoretic aggregative games and illustrate our results on an electricity market game.
查看更多>>摘要：This letter studies the issue of finite-time model-free adaptive control (FT-MFAC) applied to heterogeneous nonlinear vehicular platooning systems, focusing on a data-driven approach for design and analysis. Initially, the nonlinear vehicular platooning system is transformed into an equivalent data-relationship model through the use of pseudo partial derivatives. Subsequently, an output tuning factor is employed to facilitate the concurrent tracking of both position and velocity. Then, an adaptive controller without introducing additional constraints and shifting functions is developed to ensure model-free and finite-time control of the vehicle platoon. Ultimately, the proposed method’s effectiveness and advantages are validated through theoretical analysis and simulation results.
查看更多>>摘要：This letter investigates the moving target fencing problem for second-order multi-agent systems, where an unknown time-varying acceleration actuates the target. Without predefined distances, we propose a distributed label-free target-fencing strategy comprising a finite-time target motion estimator and a label-free target fencing controller. Relying solely on the target position, the estimator ensures the finite-time convergence of the estimation errors regarding target velocity and acceleration. The controller is designed with an attractive term that propels agents towards the target and an inter-agent repulsion term. The target is asymptotically fenced under the proposed fencing strategy, ensuring no inter-agent collisions and achieving velocity convergence, even with unexpected misbehaving agents. The results are proved by rigorous theoretical analysis and verified by numerical simulations.
查看更多>>摘要：This letter focuses on the resilient consensus problem of second-order sampled-data multi-agent systems (MASs) in the presence of mobile malicious agents under a directed graph. Unlike static attacks, mobile adversaries have the capability to move within the network, causing the state values of attacked agents to remain compromised for a certain period. To this end, we propose a modified version of the double-integrator position-based mean sub-sequence reduced (DP-MSR) algorithm, and develop conditions on the network structure to guarantee the resilient consensus of the considered system. In the above scheme, the extreme values from both neighbors and the agent itself are ignored to better compensate the impact of mobile malicious agents. It is demonstrated that the system can achieve resilient consensus under sufficient network connectivity, even in the presence of a limited number of mobile malicious agents. Numerical simulations are performed to verify the effectiveness of the proposed algorithm.
查看更多>>摘要：This letter investigates the state estimation problem under binary sensors with inaccurate thresholds. A method for extracting zonotopic data from an inaccurate threshold model is proposed. The output of this method serves as measurements to construct a centralized zonotopic Kalman Filter (ZKF). By analyzing the characteristics of binary sensors, we further propose a threshold estimation technique to encapsulate the actual threshold within a zonotope. We demonstrate that the captured zonotope becomes increasingly tighter, thereby reducing the uncertainty of the threshold estimation. Additionally, through comparisons of estimation error bounds, we extend the proposed method to the sensor arrangement problem, addressing how to select the number of sensors and their thresholds. Circuit simulations validate the effectiveness of the proposed approach.
NETL
NSTL
IEEE