Autonomous motion planning (AMP) of unmanned aerial vehicles (UAVs) is aimed at enabling a UAV to safely fly to the target without human intervention. Recently, several emerging deep reinforcement learning (DRL) methods have been employed to address the AMP problem in some simplified environments, and these methods have yielded good results. This paper proposes a multiple experience pools (MEPs) framework leveraging human expert experiences for DRL to speed up the learning process. Based on the deep deterministic policy gradient (DDPG) algorithm, a MEP–DDPG algorithm was designed using model predictive control and simulated annealing to generate expert experiences. On applying this algorithm to a complex unknown simulation environment constructed based on the parameters of the real UAV, the training experiment results showed that the novel DRL algorithm resulted in a performance improvement exceeding 20% as compared with the state-of-the-art DDPG. The results of the experimental testing indicate that UAVs trained using MEP–DDPG can stably complete a variety of tasks in complex, unknown environments.

Pointing error is a critical performance metric for vehicle-mounted single-photon ranging theodolites (VSRTs). Achieving high-precision pointing through processing and adjustment can incur significant costs. In this study, we propose a cost-effective digital correction method based on a piecewise linear regression model to mitigate this issue. Firstly, we introduce the structure of a VSRT and conduct a comprehensive analysis of the factors influencing its pointing error. Subsequently, we develop a physically meaningful piecewise linear regression model that is both physically meaningful and capable of accurately estimating the pointing error. We then calculate and evaluate the regression equation to ensure its effectiveness. Finally, we successfully apply the proposed method to correct the pointing error. The efficacy of our approach has been substantiated through dynamic accuracy testing of a 450 mm optical aperture VSRT. The findings illustrate that our regression model diminishes the root mean square (RMS) value of VSRT’s pointing error from 17″ to below 5″. Following correction utilizing this regression model, the pointing error of VSRT can be notably enhanced to the arc-second precision level.

For photoelectric theodolite, the mirror is the core optical component, so it is of great significance to design and optimize a mirror with excellent overall performance. In order to comprehensively consider the contradictory objectives of mass, natural frequency, and RMS under gravity, a multiobjective optimization method based on the performance analysis of two-parameter coupling was proposed. On the basis of the performance law, a suitable solution for balancing multiple objective functions is obtained by introducing manual intervention. The results show that compared with the traditional empirical design of mirrors, the first-order natural frequency, mass, and RMS performance of the optimized mirror are improved by 18.64%, 0.1%, and 15.58%, respectively. The frequency/Mass ratio and 1/(RMS*Mass) ratio are increased by 18.72% and 18.59%, respectively. Its comprehensive performance has been improved. This method is effective and provides a reference for the design of photoelectric theodolite and other mirrors.

This article presents a fabrication strategy on the structural design, optimization, additive manufacturing, and processing of metal mirror. Specifically, the study showcases the topology design of a metal mirror with diameter of 200 mm, the additive manufacturing of standard aluminum-based powder (AlSi10Mg), the high-precision single-point diamond turning process of the surface. By setting the feasible domain partition, a topology optimization model suitable for metal additive manufacturing and subsequent surface shaping was constructed, which takes into account the multi-load machining load conditions of single-point diamond turning technology and the material topology representation of standard support structures for additive manufacturing. The results demonstrate that the optimization model effectively suppresses the vibration phenomenon during single-point cutting. Furthermore, the results of the optical interferometer surface inspection confirm that the design and processing strategy for additively manufactured metal mirrors demonstrated in this study can be directly applied to infrared band reflective imaging optical systems.

In this paper, the experimental modes of a large-scale photoelectric theodolite tracking frame are presented. On the basis of the experimental data and the gradient-less optimization approach, the modeling strategy and the parameterized equivalent dynamic finite element model are presented. Shafting, three-point leveling units, and other components are reasonably simplified during the modeling process. Influence factors such as contact stiffness are introduced as dynamic parameters in the model. The optimized parametric model obtained demonstrates that the linearization modeling strategy represents the dynamic response characteristics of this type of structure accurately. The maximum relative error of the first four-order natural frequencies between numerical simulation and experimental data is 4.45% when the consistency of mode shapes is taken into account. The research results in this paper can provide engineering guidance for the dynamic stiffness optimization design of the large-scale photoelectric theodolite tracking frame.