The artificial potential field method is a highly popular obstacle avoidance algorithm which is widely used in the field of industrial robotics due to its high efficiency. However, the traditional artificial potential field method has poor real-time performance, making it less suitable for modern factory work patterns, and it is difficult to handle situations when the robotic arm encounters singular configurations. In this paper, we propose an improved artificial potential field method in joint space, which effectively improves the real-time performance of the algorithm, and still performs well when the robotic arm falls into a singular configuration. This method solves the gradient of the repulsive potential field in advance by defining the shortest distance from each joint of the robotic arm to the obstacle, and only needs to calculate potential field function once per cycle, which significantly reduces the calculation time. In addition, when a robotic arm falls into a local minimum position in potential field, the algorithm adds a virtual obstacle to make it leave the position, while this virtual obstacle does not require additional input information. Experimental results show that the algorithm obtains short movement paths and requires very little computing time in the face of different obstacles.
In recent years, physics-informed neural networks (PINNs) have developed significantly as a deep learning technology. In analogy to the selection of grid cells in traditional numerical methods, the distribution of sample points used for training PINN can have a greater impact on the solution accuracy. Based on the Residual-based Adaptive Refinement (RAR) algorithm, many improved adaptive sampling algorithms have been proposed. However, these sampling algorithms rely on residuals as the error indicator and focus only on the sample points within the solution domain. Therefore, we introduce a novel adaptive sampling algorithm EI-RAR. This algorithm incorporates a new expected improvement (EI) function, which increases focus on the sample points at the boundaries of the solution domain. Additionally, EI-RAR integrates attention mechanism with a sample point generation algorithm, aimed at reinforcing the connection between newly-added and existing sample points. To increase the accuracy of solving problems with sharp solutions, we build upon the EI-RAR algorithm by incorporating gradient information of the residual values as a criterion for sample point selection, leading to the development of a second adaptive sampling algorithm EI-Grad. We select residual neural network and combine it with adaptive sampling algorithms for a series of numerical experiments, aiming to reduce the phenomenon of gradient vanishing during the training process. These experiments select the Diffusion equation, Burgers’ equation, Allen-Cahn equation, and Navier-Stokes equation, respectively. Numerical results indicate that, with the same number of residual points, the EI-RAR algorithm is more precise compared to other sampling methods, and the EI-Grad algorithm can also effectively solve partial differential equations with sharp solutions.
The positioning and velocity accuracy of spraying robots determine the quality of the coating, and the influence of the robotic dynamic characteristics on control precision is significant. This paper presents a method of linearizing dynamic characteristics into feedforward coefficients and designs a dual-loop control system consisting of an inner velocity loop and an outer position loop. The system is divided into three sections: a cascaded section, a feedback section, and a feedforward section. The cascaded section eliminates the nonlinear characteristics of the system; the feedback section ensures the stability of the system; the feedforward section compensates for the internal errors of the system. The main innovation of this paper lies in proposing an offline parameter tuning method, which avoids online parameter adjustments and significantly enhances the real-time performance of the control system. Additionally, this method does not require specific physical information of the system, thus avoiding the cumbersome process of parameter adjustment. The experimental results demonstrate that when facing different high-speed trajectories, the proposed control system exhibits a significant improvement in control accuracy compared to other advanced control schemes.
Physics-informed neural networks (PINNs) have garnered widespread use for solving a variety of complex partial differential equations (PDEs). Nevertheless, when addressing certain specific problem types, traditional sampling algorithms still reveal deficiencies in efficiency and precision. In response, this paper builds upon the progress of adaptive sampling techniques, addressing the inadequacy of existing algorithms to fully leverage the spatial location information of sample points, and introduces an innovative adaptive sampling method. This approach incorporates the Dual Inverse Distance Weighting (DIDW) algorithm, embedding the spatial characteristics of sampling points within the probability sampling process. Furthermore, it introduces reward factors derived from reinforcement learning principles to dynamically refine the probability sampling formula. This strategy more effectively captures the essential characteristics of PDEs with each iteration. We utilize sparsely connected networks and have adjusted the sampling process, which has proven to effectively reduce the training time. In numerical experiments on fluid mechanics problems, such as the two-dimensional Burgers’ equation with sharp solutions, pipe flow, flow around a circular cylinder, lid-driven cavity flow, and Kovasznay flow, our proposed adaptive sampling algorithm markedly enhances accuracy over conventional PINN methods, validating the algorithm’s efficacy.
High-precision control of robotic manipulators plays a vital role in improving the efficiency and quality of industrial manufacturing. However, the inherent nonlinear and time-varying characteristics of robotic systems make high-accuracy control a challenging task, and external noise interference further complicates reliable state estimation. Conventional time-delay control methods often involve computationally intensive procedures for gain determination and are limited in their ability to suppress noise effectively. To overcome these limitations, this paper proposes a robust time-delay control strategy based on an improved state observer. By deriving a linearized form of the dynamic model, an offline computation scheme for control gain determination is developed, which eliminates the need for additional tuning parameters and simplifies the design process. Furthermore, the proposed state observer integrates model reference estimation with noise suppression techniques, enabling accurate acquisition of joint states and improving system robustness under noisy conditions. Experimental results validate the effectiveness of the proposed method, showing that it can efficiently determine control gains and significantly outperform existing advanced approaches in terms of trajectory tracking accuracy and overall control performance.