RRT-Based Trajectory Planning And Control For Spacecraft Autonomous Rendezvous And Proximity Operations

In recent years,spacecraft relative guidance and control technology has attracted extensive attention due to its application requirements in complex missions such as ren-dezvous and proximity operations,space surveillance and asteroid landing&sampling.For non-cooperative space target proximity operations with high-dimension state,complex constraints,dynamic environment,system uncertainties,designing a feasible guidance and control(G&C)system while considering real-time feasibility,optimality and robustness is of great importance and difficulty nowadays.Based on sampling-based motion planning(SBMP)and receding horizon control(RHC),this paper further studies the G&C problem of tracking a target spacecraft in close range.And this paper mainly includes:According to the characteristics of spacecraft rendezvous and proximity operations,a framework of the optimal trajectory planning problem is established,and the basic knowledge of SBMP algorithm is discussed,which lays a foundation for further research.Firstly,the relative motion trajectory dynamics,rigid body attitude dynamics,collision avoidance,image sensor pointing,thruster plume,control feasibility and terminal condi-tions are described using mathematical language.Then,the basic principle,procedures and properties of the optimal rapidly-exploring random tree(RRT~*)algorithm is in-troduced.Finally,considering a cooperative space target,the spacecraft translational rendezvous trajectory planner based on RRT~*algorithm is designed continuous thrust modes.Since the cooperative target can actively provide its state information,the con-straints are relatively simple,including only linear relative motion dynamics,collision avoidance,thruster plume and control input constraints.In the numerical simulation,the chaser successfully achieves short-range flying around the cooperative target,which proves that SBMP method is working in this context,but the runtime is still not able to meet the requirement of real-time implementability.An adaptive sampling strategy based on learning is designed to improve the con-ventional rapidly-exploring random tree(RRT)algorithm.Through analysis,it is found that the sampling strategy has a great impact on the computational efficiency of SBMP algorithm,but the conventional RRT algorithm fails to use the information acquired in the exploration process to improve the planning results.Therefore,an adaptive sampling strategy is designed to update the sampling space with tree node information and optimize the cost of resultant trajectory.To balance the accuracy and computational efficiency,the gaussian mixture model was used to describe the distribution of the samples,and a elite set was selected according to the cost information,and the robust EM clustering algo-rithm was used to learn the sample distribution online.Through theoretical analysis and numerical simulation experiments,the effectiveness of the proposed algorithm is verified,and the convergence of the algorithm is accelerated with probability completeness and asymptotic optimality.For space non-cooperative target,the optimal trajectory planning problem of the integrated position and attitude dynamical system is established considering the chaser implementing active detection system to obtain the target state,and solved by the adaptive RRT~*algorithm.The problem includes a lot of constraints,such as integrated pose dy-namics,collision avoidance,visual sensor orientation,thruster plume,control feasibility,and docking capture terminal conditions,etc.Firstly,considering the requirement of on-line computation,the local two-point boundary value problem is solved using polynomial trajectory parameterization method.Then,the stability of deterministic sampling and the high efficiency of random adaptive sampling are combined to generate a 12-dimensional motion state.Then,a cost measure is used instead of the Euclidean distance to calculate the neighbors,and the position and attitude cost reachable set are described to facilitate parameter tuning.Finally,based on the principle of linear superposition,a trajectory smoothing algorithm is designed to improve the quality of the trajectory.The effec-tiveness of the algorithm is verified by numerical simulation,and the influence of the planner parameters and initial values on the algorithm is further discussed.Compared with the pseudo-spectral method,the sampling motion planning algorithm is able to return a near-optimal trajectory while keeping the real-time performance.A fast online replanning algorithm based on time horizon is designed to solve the problems with obstacle constraint appear,disappear and move in dynamic environment.Considering inertia uncertainty,space perturbation,measurement noise and other uncer-tainties,as well as the influence of uncontrolled tumbling non-cooperative target on the performance,an online TRH-RRT~*guidance&control algorithm was designed com-bining sampling-based motion replanning and RHC method.Secondly,combined with the proposed HL-RRT~*algorithm,the open-loop trajectory planner is designed,and the sub-problem of optimization in each guidance cycle is established.Then the error pose dynamics model is established and a non-singular terminal sliding mode control method is proposed for fast and stable tracking of pose trajectory.Finally,the stability and recursion feasibility of the proposed guidance control algorithm are analyzed,and the effectiveness of the algorithm is verified by two close operation examples of spacecraft.Numerical simulation results show that the sampling-based motion planning algorithm is capable of solving the integrated position and attitude synchronous planning problem under complex and multiple constraints quickly and stably,and can address the problems with strong coupling,strong nonlinear and strong constraints caused by uncontrolled tumbling non-cooperative target,which shows the potential in engineering application.