论文部分内容阅读
针对基于视觉导航的航迹规划技术存在的两个问题:第一,作为一类存在有限测量量程的导航系统,视觉导航系统需要目标始终出现在摄像头的视野内;第二,当目标始终在摄像头的光轴延长线附近时,视觉算法能够更有效率并且更准确的解决位姿估计问题;文中首先将航迹优化问题转化为最优控制问题,结合飞行器运动动力学特性,设计最优控制问题的约束与最优指标。然后,文中利用伪谱法将连续时间最优控制问题转化为离散时间非线性规划问题,并在此基础上研究了基于滚动时域的动态规划方法,最终减少视觉导航误差对航迹优化的不良影响。
There are two problems in the trajectory planning technology based on visual navigation. First, as a kind of navigation system with limited measuring range, the visual navigation system needs the target to always appear in the camera’s field of vision. Second, when the target is always in the camera , The visual algorithm can solve the pose estimation problem more efficiently and accurately. In this paper, the trajectory optimization problem is first transformed into the optimal control problem. Combining with the kinematic characteristics of the aircraft, the optimal control problem Constraints and Optimal Indexes. Then, by using the pseudospectral method, the problem of continuous time optimal control is transformed into a discrete time nonlinear programming problem. On the basis of this, the dynamic programming method based on rolling time domain is studied to reduce the adverse effect of visual navigation error on trajectory optimization influences.