In this paper, we address a decision making model of evacuation behavior. First, we introduce an individual evacuation model with conformity bias. Next, the evacuation model is expanded to the one as population. By focusing on the statics, we compare the conformity bias model with a previous one dealt with in general population decision making, and investigate similarities and differences between these two models. In addition, we clarify the impact of the conformity bias to dynamic evacuation behavior in terms of passivity. Finally, we present a passivity-based nudging mechanism for the evacuation behavior.

Control Barrier Functions (CBFs) have attracted much attention in recent years as a control method to guarantee safety. Previous studies have shown the effectiveness of obstacles and robots of various shapes. However, there is no study using a CBF that considers the shape of the target system and the obstacle simultaneously. In this paper, we derive a control law of considering the shape of the target system and design CBF for non-convex safe sets. Then, we construct a human assist input by using the CBF. Moreover, we confirm the effectiveness of the proposed method by experiments of an arm robot.

This paper discusses a data-driven control method. In particular, we focus on Estimated Response Iterative Tuning (ERIT) which updates the feedforward controller directly from data. This paper employs impulse response representation for feedforward controller to make the freedom sufficiently large. We also proposes to use kernel regularization to avoid overfitting. The effectiveness of the proposed method is shown through numerical example and practical experiments.

This paper is concerned with sequential distributed development of multiple retrofit controllers that are managed by respective controller designers. Retrofit control is a modular control approach for stable network systems whose subsystems are controlled by their corresponding local controllers. From the standpoint of a single controller designer, the subsystems managed by all other designers can be regarded as an unknown environment, an approximate model of which is supposed to be accessible in local controller design. This paper shows that, for the identification of approximate environment models, each of all environments for controller designers are invariant with respective to all design parameters in retrofit controllers. Furthermore, it is also shown that, for the entire feedback system, control performance regarding a local evaluation in response to a local disturbance is improved as improving the accuracy of a corresponding approximate environment model, while control performance regarding a global evaluation is not in general. The significance of the theoretical results is verified through an example of power systems control.

In this paper, the regularized and smoothed Fischer-Burmeister (FBRS) method is introduced in model predictive control based on the continuation method. By introducing the FBRS method, an optimal control problem formed as a convex problem was regularized and smoothed to satisfy regularity of an equation derived by using a continuation method. In addition, the equation was reduced to a general linear system and a symmetric positive definite system to minimize the computational burden. Furthermore, we discussed the effect of the regularizing and the smoothing parameters on condition number related to numerical accuracy in computation. A guiding principle for the parameters was obtained via the discussion. Finally, an example for a constrained multi-input multi-output system, control design of an engine airpath system was demonstrated to confirm the reasonability of the method.

The idea of the model predictive control (MPC) is applied to a space launch vehicle attitude control in order to deal with two major issues of this field: the uncertain and time-varying characteristic of the dynamics. The rocket's parameters such as the mass and the stiffness are constantly changing with time and have uncertainty because it is not possible to test a flight model. Additionally, this study is focused on attitude control of the launch vehicle during its ascent stage. At this stage, the environmental condition is also changing with time, and there is uncertainty in local wind and atmospheric conditions. The latest designs, including a combination of H_∞ control and gain scheduling control, have been successfully applied to real systems. The gain scheduling control approach divides the entire flight time into multiple blocks. This approach requires significant time and labor because each block must have its own controller design. To simplify this design procedure, the goal of the study is to design a robust control algorithm that can directly respond to the time-varying characteristics of the system. In this paper, two controllers are designed: one for the time-varying system and the other for the uncertain system. The one has been designed with adaptive model predictive control (AMPC). AMPC can predict future behavior of the system more precisely than conventional MPC. The other controller uses the idea of linear matrix inequality (LMI) for robust MPC (RMPC). Through simulations, it is shown that AMPC is able to deal with time variations in dynamics and RMPC has robust stability.

In this paper, we present a new method of real-time tuning by directly using experimental data. Here, we focus on I-PD controller which is widely used in many actual control applications. We also expand Virtual Internal Model Tuning, which was proposed by the authors and is an off-line tuning method of controller parameter by using directly experimental data, to the real-time manner. Particularly, we show that our proposed real-time tuning converges to the optimal parameter from the theoretical points of view. Finally, numerical examples is illustrated to show the validity of our theoretical analysis on the convergence.

In conventional model-based controls, the control performance depends on the accuracy of models used in the controller design. Therefore, in the case where the modeling errors occur due to changes in environmental conditions and/or aging deterioration, the control performance is deteriorated. In this paper, a control system design strategy based on “smart model-based development (S-MBD)” is presented. Smart MBD is a concept of a control system design approach using models and data. In the smart MBD approach, introducing a data-driven controller into the model-based control system can prevent the degradation of the control performance caused by the model uncertainties. The effectiveness of the proposed control method is confirmed by numerical examples on a vehicle yaw rate tracking control problem.

Space missions require “resilience” to flexibly complete the mission in response to changes in the environment and system characteristics. The present study proposes a method for autonomously planning a corrective control law for lunar landing trajectory control to cope with off-nominal conditions and reflecting it in resilience improvement measures by utilizing reinforcement learning. The proposed method employs a reinforcement learning problem in which an agent is additionally placed in the control loop and the corrective control input as an action output by the agent is added to the original closed-loop control input. The results and insights are summarized for the resultant agent's characteristics which autonomously detect off-nominal conditions and proactively implement recovery measures, while verifying the capability and effectiveness of the proposed design framework enabled by a reinforcement learning architecture in a realistic and specific lunar landing sequence.