Deep reinforcement learning for six degree-of-freedom planetary landing

This work develops a deep reinforcement learning based approach for Six Degree-of-Freedom (DOF) planetary powered descent and landing. Future Mars missions will require advanced guidance, navigation, and control algorithms for the powered descent phase to target specific surface locations and achieve pinpoint accuracy (landing error ellipse <5 m radius). This requires both a navigation system capable of estimating the lander’s state in real-time and a guidance and control system that can map the estimated lander state to a commanded thrust for each lander engine. In this paper, we present a novel integrated guidance and control algorithm designed by applying the principles of reinforcement learning theory. The latter is used to learn a policy mapping the lander’s estimated state directly to a commanded thrust for each engine, resulting in accurate and almost fuel-optimal trajectories over a realistic deployment ellipse. Specifically, we use proximal policy optimization, a policy gradient method, to learn the policy. Another contribution of this paper is the use of different discount rates for terminal and shaping rewards, which significantly enhances optimization performance. We present simulation results demonstrating the guidance and control system’s performance in a 6-DOF simulation environment and demonstrate robustness to noise and system parameter uncertainty.     

Autonomous orbit determination using epoch-differenced gravity gradients and starlight refraction

Autonomous orbit determination via integration of epoch-differenced gravity gradients and starlight refraction is proposed in this paper for low-Earth-orbiting satellites operating in GPS-denied environments.Starlight refraction compensates for the significant along-track position error that occurs from only using gravity gradients and benefits from integration in terms of improved accuracy in radial and cross-track position estimates.The between-epoch differencing of gravity gradients is employed to eliminate slowly varying measurement biases and noise near the orbit revolution frequency.The refraction angle measurements are directly used and its Jacobian matrix derived from an implicit observation equation.An information fusion filter based on a sequential extended Kalman filter is developed for the orbit determination.Truth-model simulations are used to test the performance of the algorithm,and the effects of differencing intervals and orbital heights are analyzed.A semi-simulation study using actual gravity gradient data from the Gravity field and steady-state Ocean Circulation Explorer (GOCE) combined with simulated starlight refraction measurements is further conducted,and a three-dimensional position accuracy of better than 100 m is achieved.     

Tube-based robust output feedback model predictive control for autonomous rendezvous and docking with a tumbling target

In this paper, a tube-based robust output feedback model predictive control method (TRMPC) is proposed for controlling chaser spacecraft docking with a tumbling target in near-circular orbit. The controller contains a simple, stable, Luenberger state estimator and a tube-based robust model predictive controller. Several practical challenges are also considered under dock-enabling conditions, such as the control saturation, velocity constraint, approach corridor constraint, and collision avoidance constraint. Meanwhile, uncertainties are carefully analyzed when designing the controller, including dynamics uncertainty, measurement error, and control deviation. The TRMPC ensures that all possible state trajectories with uncertainties lie in the minimum robust positively invariant set (mRPI, i.e., the so-called tube in this paper). The tube center is the solution of a nominal (without uncertainties) system. Another important contribution of this paper is to propose a technique where it is unnecessary to calculate the mRPI explicitly. Thereby, the ‘curse of dimensionality’ can be avoided for a six-dimensional system. To verify the feasibility of the proposed TRMPC strategy in the presence of uncertainties, two scenarios of autonomous rendezvous and docking (AR&D) are simulated. The simulation results show that the TRMPC method is more efficient in minimizing the uncertainties, fuel consumption, and computational cost, compared to the classic model predictive control (MPC) method.     

基于CAN总线的全自主移动车辆信息网络设计

面向全自主轮式移动车辆设计了基于CAN总线的控制-感知-导航通信系统。提出了CAN总线带宽分配原则及通信协议,并针对典型导航传感器系统进行了通信负载-可靠性试验,给出了多自主移动车辆的信息网络的建立方法。实验结果表明:基于CAN总线实现车体导航-控制可满足自主控制的实时性和可靠性要求。     

基于载波相位的完整性监测及其优化方法

根据垂直保护限准则,利用载波相位观测信息,提出一种用于飞机精密进近着陆阶段的GPS接收机自主完整性监测优化方法。该方法不增加外部观测量,将整个飞机精密进近着陆过程分成若干独立观测时段,时段中任意一个历元报警,则整个时段报警;在满足国际民航组织规定的每个时段误警率和漏检率设计指标前提下,时段误警事件是单历元误警事件的和事件,时段漏检事件是单历元漏检事件的积事件。试验结果表明,应用该优化方法,在垂直方向上,漏检概率由时段的1×10~(-9)/15s增加到单历元的1×10~(-0.6)/s,有效地减小了垂直保护限值VPL,增强了GPS卫星导航系统的可用性。     

导航星座不完整对星间自主定轨几何精度因子的影响

星间自主定轨是星座自主导航的关键技术. 在系统建设初期未完成全星座组网、卫星出现故障或受损等情况下, 部分卫星缺失将导致导航星座不完整, 其对星间自主定轨性能的影响值得研究. 本文在提出逐卫星加权最小二乘自主定轨估计方法的基础上, 引入几何精度因子作为衡量星座不完整影响卫星自主定轨性能的指标, 最后以Galileo星座为例进行了仿真与分析. 结果表明, Galileo星座中卫星进行自主定轨时其可视卫星的冗余度较高, 少数卫星缺失不会对星间自主定轨的几何精度因子产生明显影响. 只有当星座缺失卫星数达2/3时, 会使得部分卫星的几何精度因子超差, 卫星自主定轨性能明显下降.      

连续小推力航天器自主导航及在轨参数标定

以连续小推力航天器为背景,提出了综合考虑星载加速度计和推力器在轨标定的自主导航方案。首先以精确姿态测量和引力梯度模型为标定参考信息源,建立了包含加速度计参数、推力器参数以及光压系数的完整参数测量模型;然后基于天文导航方法建立了自主导航系统状态模型和观测模型;表明各状态和参数的能观性后,采用了具备良好计算效率和鲁棒性的双重无迹卡尔曼滤波方法进行状态和参数联合估计。分析与数值仿真表明,该方法通过结合参数在轨标定直接提高了导航模型精度,在工程应用中具备可行性和有效性。     

考虑月球扁率修正的月球卫星自主导航

针对月球扁率对月球紫外敏感器的月心方向矢量确定的不利影响,研究了月球紫外敏感器的测量原理和敏感到月平边缘时满足的几何约束,提出了一种考虑月球扁率的月心矢量确定方法。并进一步的结合地球敏感器和太阳敏感器的测量信息,研究了基于日地月方位信息的月球卫星自主轨道算法,并评估了月球方位确定算法对导航精度的影响。仿真结果表明,在太阳敏感器、地球敏感器和月球敏感器的精度分别为0.02°(3σ)、 0.05°(3σ) 和0.1°(3σ)的假设下,考虑月球扁率修正的月球卫星的自主导航位置精度能达到300m(3σ),导航速度误差能达到0.6m/s(3σ), 从而保证了环月卫星的导航精度。     

星光折射导航星的改进三角识别捕获

高精度星光折射间接敏感地平的自主导航方法,目前被国内外尤为关注。本文通过对星光折射导航观测对象———导航星捕获的研究,采用了改进的三角匹配算法并结合恒星星表,确定出导航观测星星光方向,同时依据对飞行器在轨观测星的几何关系推导,利用Unscented卡尔曼滤波算法确定航天器的在轨位置和速度。该方法能比较真实地模拟飞行器实际在轨的导航情况。仿真表明,真实星表导航精度大于人造星场;通过比较导航系统捕获20,40,60,80颗折射星的4种情况,可知捕获到的导航星的数量越多,导航精度也随之提高。     

深空探测低信噪比通信研究

与传统无线通信相比, 深空通信由于传输损耗巨大、探测器能量受限且轨道多变等因素, 须解决信噪比(Signal to Noise Ratio, SNR)极低、数据接收不连续及信道参数多变等难题. 针对这些问题, 论述了在深空探测极低SNR下需重点研究的深空天线组阵、编码与调制、联合解调译码、喷泉码以及其自主无线电实现等关键技术, 以有效协助解决上述深空通信难题. 同时, 讨论了这些关键技术的发展趋势及其在未来深空通信中的应用.