In this paper, a model predictive control (MPC) strategy is developed to realize collision avoidance with dynamic obstacle vehicles for an autonomous vehicle. A novel collision constraint based on road width and vehicle shape is proposed to ensure a safe distance between the autonomous vehicle and an obstacle vehicle. A smooth path is guaranteed in collision avoidance by the adaptive artificial potential field (APF) method for the autonomous vehicle using the MPC strategy. Both recursive feasibility and practical stability are analyzed for the autonomous vehicle based on the MPC strategy with multiple constraints and variable longitudinal velocity. Experimental results are presented to demonstrate the effectiveness and superiority of the MPC strategy with the adaptive APF method.
6 June 2023 in IEEE Transactions on Systems, Man, and Cybernetics: Systems
In this article, the distributed model predictive control (MPC)-based noncooperative game problem is dealt with for the discrete-time multiplayer systems (MPSs) with an undirected graph. To reflect the reality, the state and input constraints are considered along with the matched disturbances and unmatched disturbances. The disturbance-observer-based composite MPC strategy is put forward which optimizes a given cost function over the receding horizon while eliminating the matched disturbances. An iterative algorithm is developed such that the model predictive dynamic game (MPDG) converges to the so-called $\varepsilon $ -Nash equilibrium in a distributed manner. Sufficient conditions are established to guarantee the convergence of the proposed algorithm. In addition, easy-to-check conditions are also provided to ensure the uniform boundedness of the studied MPSs. Finally, a numerical example of a group of spacecrafts is provided to verify the effectiveness of the proposed methodology.
In decentralized tracking systems, matching and fusion of target tracks in local sensors are key to improve estimation performance. However, sensor measurements usually contain bias and clutter due to sensor performance or environmental influence, which makes it difficult to match local tracks correctly and obtain accurate tracking results. Moreover, the mutual influence of local track matching and sensor bias estimation further aggravates the difficulty. This paper proposed a joint matching and fusion optimization framework to address the decentralized fusion problem for multi-target tracking systems with sensor bias and clutter. To deal with the impact of clutter, a direct relationship between local estimates and bias is obtained based on the joint probabilistic data association (JPDA) filter. A hypothesis test is applied in matching detection of local tracks, the target matching results at relatively sparse locations are obtained with less computational cost. Then soft matching and fusion estimation are iteratively implemented to gradually adjust the matching results and sensor bias estimates, in which Kullback-Leibler (KL) distance is introduced to better measure the similarity between two local tracks. Finally, simulation results of multi-target tracking are provided to verify that the proposed method can obtain more accurate fusion and bias estimates and has lower computational complexity compared to other methods.
23 May 2025 in IEEE Transactions on Automation Science and Engineering
In this paper, a fusion estimator is investigated for positioning systems with location constraint of a camera and an ultra-wideband (UWB). Local camera and UWB estimators are designed independently for positioning with missed detections and local UWB estimates are aligned to local camera estimates for fusion estimation. Next the location constraint of the camera and UWB is transformed to a state constraint by state extension, and a constrained estimation projection method is applied to construct a fusion estimator. To measure the accuracy of the designed constrained fusion estimator, it is demonstrated that the estimation error covariance of the proposed method is equal to that of the matrix-weighted fusion method. Moreover, it is found that the proposed method has significant advantages over the matrix-weighted fusion method in computational cost, which is crucial for the real-time requirements of perception systems. Finally, the effectiveness of the proposed fusion estimator is displayed by simulation results and experimental results on the positioning system for an Ackerman vehicle. Note to Practitioners—The proposed method aims to solve the challenges of missed detections and different location coordinates from two sensors in the positioning systems. The paper introduces independently designed local estimators for both the camera and UWB sensors, which are capable of handling missed detections. The local UWB estimates are aligned with the local camera estimates to ensure that local estimates from two sensors are time-aligned in the fusion process. The location constraint of the sensors is transformed into a state constraint through state extension for constructing the fusion estimator with the constrained estimation projection method. The paper demonstrates that the estimation error covariance of the proposed fusion estimator is equivalent to that of the traditional matrix-weighted fusion method. Moreover, the proposed method offers significant advantages in terms of computational cost, which is critical for real-time performance in perception systems. Practitioners can use the proposed fusion estimator to enhance the accuracy and reliability of positioning systems that incorporate two different sensors with missed detections and location constraint. The method is particularly well-suited for applications where real-time performance is a requirement.
In this paper, we explore the problem of smooth braking at preset points in autonomous vehicles using model predictive control (MPC) with a receding horizon extended state observer (RHESO) and a neural network (NN). An NN-based modeling method is proposed to intuitively describe the relationship between vehicle speed and the vehicle controllers (brake and throttle), and establish a dynamic model of autonomous vehicles. A sufficient condition is put forward to guarantee the convergence of the proposed NN. Furthermore, a composite MPC strategy based on RHESO is designed, which optimizes a given cost function over the receding horizon while mitigating the effects of modeling inaccuracies and disturbances. Additionally, easily verifiable conditions are provided to ensure the autonomous driving vehicles’ uniform boundedness. Numerically illustrative examples are given to demonstrate the effectiveness of the proposed approach.
This paper presents a model predictive control (MPC)-enabled disturbance-rejection controller approach for a pneumatic manipulator system subjected to complex nonlinear terms within the system. To facilitate the handling of the complex nonlinear terms, they are modeled as disturbances. To address these disturbances, a receding-horizon-based extended state observer (RH-ESO) incorporating a decision variable is developed. The optimal disturbance estimation error is determined through a receding-horizon optimization procedure, which provides the best estimate of the disturbance. Using this optimal estimate, the MPC-enabled disturbance-rejection controller is proposed for the pneumatic manipulator system to achieve angle tracking control. Moreover, the proposed approach ensures both the recursive feasibility of the optimization problem and the uniform boundedness of the closed-loop system. The simulation results further demonstrate the effectiveness and validity of the proposed methodology.