For the locomotion control of a legged robot, both model predictive control (MPC) and reinforcement learning (RL) demonstrate powerful capabilities. MPC transfers the high-level task to the lower-level joint control based on the understanding of the robot and environment, model-free RL learns how to work through trial and error, and has the ability to evolve based on historical data. In this work, we proposed a novel framework to integrate the advantages of MPC and RL, we learned a policy for automatically choosing parameters for MPC. Unlike the end-to-end RL applications for control, our method does not need massive sampling data for training. Compared with the fixed parameters MPC, the learned MPC exhibits better locomotion performance and stability. The presented framework provides a new choice for improving the performance of traditional control.

This paper focuses on the design of a disturbance rejection controller for a tailless aircraft based on the technique of nonlinear dynamic inversion (NDI). The tailless aircraft model mounted on a three degree-of-freedom (3-DOF) dynamic rig in the wind tunnel is modeled as a nonlinear affine system subject to mismatched disturbances. First of all, a baseline NDI attitude controller is designed for sufficient stability and good reference tracking performance of the nominal system. Then, a nonlinear disturbance observer (NDO) is supplemented to the baseline NDI controller to estimate the lumped disturbances for compensation, including unmodeled dynamics, parameter uncertainties, and external disturbances. Mathematical analysis demonstrates the convergence of the employed NDO and the resulting closed-loop system. Furthermore, an anti-windup modification is applied to the NDO for control performance preserving in the presence of actuator saturation. Subsequently, the designed control schemes are preliminarily validated and compared via simulations. The baseline NDI controller demonstrates satisfactory attitude tracking performance in the case of nominal simulation; the NDO augmented NDI controller presents significantly improved ability of disturbance rejection when compared with the baseline NDI controller in the case of robust simulation; the anti-windup modified scheme, rather than the baseline NDI controller nor the NDO augmented NDI controller, can preserve the closed-loop performance in the case of actuator saturation. Finally, the baseline NDI scheme and the NDO augmented NDI scheme are implemented and further validated in the wind tunnel flight tests, which demonstrate that the experimental results are in good agreement with that of the simulations.

There are many undulating terrains in the wild environment. In order to realize the adaptive and stable walking of quadruped robots on unknown sloped terrain, a slope-adaptability model predictive control (SAMPC) algorithm is proposed in this work. In the absence of external vision sensors, the orientation and inclination of the slope are estimated based on the joint position sensors and inertial measurement units (IMU). In an effort to increase the stability margin, the adaptive algorithm adjusts the attitude angle and the touch-down point of the swing leg. To reduce the slipping risk, a nonlinear control law is designed to determine the friction factor of the friction cone constraint from the inclination of the slope. We validate the effectiveness of the framework through a series of simulations. The automatic smooth transition from the flat to the unknown slope is achieved, and the robot is capable of walking in all directions on the slope. Notably, with reference to the climbing modal of blue sheep, the robot successfully climbed a 42.4° slope, proving the ultimate ability of the proposed framework.

针对连续观点动力学模型中缺乏对个体信任及其观点的不同权重的考虑，本文通过扩展Hegselmann-Krause模型，引入个体间的信任度和观点间的相似度，将有限信任假设扩展为有限影响假设，建立带权重的观点更新模型，以研究社会群体中少数的偏执个体和权威个体对观点形成、演化以及共识构建过程的影响. 仿真结果表明，这两类个体的初始观点尽量接近观点分布区间中点0.5或者增加偏执个体的影响阈值和权威个体的可信任程度可以使群体形成更少和更大的观点簇. 随着偏执个体的影响阈值增加，最大簇的观点位置集中于0.5附近；而增加权威个体的可信任程度，最大簇的观点不断接近权威个体的最终观点. 模型结果能够在一定程度上反映和解释现实社会中偏执个体和权威个体现象.