Dynamic Neural Network-Based Adaptive Tracking Control for an Autonomous Underwater Vehicle Subject to Modeling and Parametric Uncertainties

Linear matrix genetic programming as a tool for data-driven black-box control-oriented modeling in conditions of limited access to training data

In the paper, a new evolutionary technique called Linear Matrix Genetic Programming (LMGP) is proposed. It is a matrix extension of Linear Genetic Programming and its application is data-driven black-box control-oriented modeling in conditions of limited access to training data. In LMGP, the model is in the form of an evolutionarily-shaped program which is a sequence of matrix operations. Since the program has a hidden state, running it for a sequence of input data has a similar effect to using well-known recurrent neural networks such as Long Short-Term Memory (LSTM) or Gated Recurrent Unit (GRU). To verify the effectiveness of the LMGP, it was compared with different types of neural networks. The task of all the compared techniques was to reproduce the behavior of a nonlinear model of an underwater vehicle. The results of the comparative tests are reported in the paper and they show that the LMGP can quickly find an effective and very simple solution to the given problem. Moreover, a detailed comparison of models, generated by LMGP and LSTM/GRU, revealed that the former are up to four times more accurate than the latter in reproducing vehicle behavior.

Trajectory Tracking and Obstacle Avoidance of Robotic Fish Based on Nonlinear Model Predictive Control

The attainment of accurate motion control for robotic fish inside intricate underwater environments continues to be a substantial obstacle within the realm of underwater robotics. This paper presents a proposed algorithm for trajectory tracking and obstacle avoidance planning in robotic fish, utilizing nonlinear model predictive control (NMPC). This methodology facilitates the implementation of optimization-based control in real-time, utilizing the present state and environmental data to effectively regulate the movements of the robotic fish with a high degree of agility. To begin with, a dynamic model of the robotic fish, incorporating accelerations, is formulated inside the framework of the world coordinate system. The last step involves providing a detailed explanation of the NMPC algorithm and developing obstacle avoidance and objective functions for the fish in water. This will enable the design of an NMPC controller that incorporates control restrictions. In order to assess the efficacy of the proposed approach, a comparative analysis is conducted between the NMPC algorithm and the pure pursuit (PP) algorithm in terms of trajectory tracking. This comparison serves to affirm the accuracy of the NMPC algorithm in effectively tracking trajectories. Moreover, a comparative analysis between the NMPC algorithm and the dynamic window approach (DWA) method in the context of obstacle avoidance planning highlights the superior resilience of the NMPC algorithm in this domain. The proposed strategy, which utilizes NMPC, demonstrates a viable alternative for achieving precise trajectory tracking and efficient obstacle avoidance planning in the context of robotic fish motion control within intricate surroundings. This method exhibits considerable potential for practical implementation and future application.

Hyperparameter Optimization for the LSTM Method of AUV Model Identification Based on Q-Learning

An accurate mathematical model is a basis for controlling and estimating the state of an Autonomous underwater vehicle (AUV) system, so how to improve its accuracy is a fundamental problem in the field of automatic control. However, AUV systems are complex, uncertain, and highly non-linear, and it is not easy to obtain through traditional modeling methods. We fit an accurate dynamic AUV model in this study using the long short-term memory (LSTM) neural network approach. As hyper-parameter values have a significant impact on LSTM performance, it is important to select the optimal combination of hyper-parameters. The present research uses the improved Q-learning reinforcement learning algorithm to achieve this aim by improving its recognition accuracy on the verification dataset. To improve the efficiency of action exploration, we improve the Q-learning algorithm and choose the optimal initial state according to the Q table in each round of learning. It can effectively avoid the ineffective exploration of the reinforcement learning agent between the poor-performing hyperparameter combinations. Finally, the experiments based on simulated or actual trial data demonstrate that the proposed model identification method can effectively predict kinematic motion data, and more importantly, the modified Q-Learning approach can optimize the network hyperparameters in the LSTM.

Synergetic Synthesis of Nonlinear Laws of Throttle Control of a Pneumatic Drive

Currently, a significant trend in control in robotic systems is developing and improving linear and nonlinear control algorithms to improve the overall quality of production with high accuracy and adaptability. The present study considers a synergistic synthesis of throttle control of a pneumatic distributor valve and backpressure control for piston rod positioning. The article presents the synthesis of control laws for the position of a pneumatic cylinder piston using the method of analytical design of aggregated regulators (ADAR) of synergetic control theory (STC), which allows operation with nonlinear mathematical models, eliminating the loss of information about the object during linearization. A comparative calculation of the energy efficiency of backpressure control and throttle control methods was carried out, while the numerical value of the total airflow with throttle control is 0.0569 m3⁄s and, with backpressure control, it is 0.0337 m3⁄s. Using a P controller in a linear model gives a transient oscillatory process damped in 2–2.5 s. When using a PID controller, the process has an overshoot equal to 11.5%, while the synergistic controller allows you to smoothly move the drive stem to a given position without overshoot. The parametric uncertainty analysis of the considered mathematical model is carried out. The model’s main parameters are identified, which change the actual functioning of the system under study. The inconsistency of applying classical control laws based on typical controllers to parametrically indeterminate mathematical models is shown.