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.

In the complex and variable marine environment, the navigation and localization of autonomous underwater vehicles (AUVs) are very important and challenging. When the conventional Kalman filter (KF) is applied to the cooperative localization of leader–follower AUVs, the outliers in the sensor observations will have a substantial adverse effect on the localization accuracy of the AUVs. Meanwhile, inaccurate noise covariance matrices may result in significant estimation errors. In this paper, we proposed an improved Sage–Husa adaptive extended Kalman filter (improved SHAEKF) for the cooperative localization of multi-AUVs. Firstly, the measurement anomalies were evaluated by calculating the Chi-square test statistics based on the innovation. The detection threshold was determined according to the confidence level of the Chi-square test, and the Chi-square test statistics exceeding the threshold were regarded as measurement abnormalities. When measurement anomalies occurred, the Sage–Husa adaptive extended Kalman filter algorithm was improved by suboptimal maximum a posterior estimation using weighted exponential fading memory, and the measurement noise covariance matrix was adjusted online. The numerical simulation of leader–follower multi-AUV cooperative localization verified the effectiveness of the improved SHAEKF and demonstrated that the average root mean square and the average standard deviation of the localization errors based on the improved SHAEKF were significantly reduced in the case of the presence of measurement abnormalities.

This paper focuses on generating distributed flocking strategies via imitation learning. The primary motivation is to improve the swarm robustness and achieve better consistency while respecting the communication constraints. This paper first proposes a quantitative metric of swarm robustness based on entropy evaluation. Then, the graph importance consistency is also proposed, which is one of the critical goals of the flocking task. Moreover, the importance-correlated directed graph convolutional networks (IDGCNs) are constructed for multidimensional feature extraction and structure-related aggregation of graph data. Next, by employing IDGCNs-based imitation learning, a distributed and scalable flocking strategy is obtained, and its performance is very close to the centralized strategy template while considering communication constraints. To speed up and simplify the training process, we train the flocking strategy with a small number of agents and set restrictions on communication. Finally, various simulation experiments are executed to verify the advantages of the obtained strategy in terms of realizing the swarm consistency and improving the swarm robustness. The results also show that the performance is well maintained while the scale of agents expands (tested with 20, 30, 40 robots).

This paper proposes an online uniform foot location planning method (UPMPC) based on model predictive control (MPC) for solving the problem of large posture changes during gait transitioning. This method converts the foot location planning into a discrete-time MPC problem. The core part of the method is to complete the planning of the foot location based on the linear inverted pendulum (LIP) model and the simplified robot dynamics model. By unifying the input foot location at each time step, the solution time is shortened. The final simulation experiment compares the results of using the UPMPC and foot location planning method with heuristic function (HF) for gait transitioning, respectively. This result demonstrates that the UPMPC can complete the gait transitioning task and adapt to large changes in posture during gait transitioning. In addition, the results also show the good performance of UPMPC in fixed gait.