The increasing complexity of on-orbit tasks imposes great demands on the flexible operation of space robotic arms, prompting the development of space robots from single-arm manipulation to multi-arm collaboration. In ...The increasing complexity of on-orbit tasks imposes great demands on the flexible operation of space robotic arms, prompting the development of space robots from single-arm manipulation to multi-arm collaboration. In this paper, a combined approach of Learning from Demonstration (LfD) and Reinforcement Learning (RL) is proposed for space multi-arm collaborative skill learning. The combination effectively resolves the trade-off between learning efficiency and feasible solution in LfD, as well as the time-consuming pursuit of the optimal solution in RL. With the prior knowledge of LfD, space robotic arms can achieve efficient guided learning in high-dimensional state-action space. Specifically, an LfD approach with Probabilistic Movement Primitives (ProMP) is firstly utilized to encode and reproduce the demonstration actions, generating a distribution as the initialization of policy. Then in the RL stage, a Relative Entropy Policy Search (REPS) algorithm modified in continuous state-action space is employed for further policy improvement. More importantly, the learned behaviors can maintain and reflect the characteristics of demonstrations. In addition, a series of supplementary policy search mechanisms are designed to accelerate the exploration process. The effectiveness of the proposed method has been verified both theoretically and experimentally. Moreover, comparisons with state-of-the-art methods have confirmed the outperformance of the approach.展开更多
The contact point configuration should be carefully chosen to ensure a stable capture,especially for the non-cooperative target capture mission using multi-armed spacecraft.In this work scenario,the contact points on ...The contact point configuration should be carefully chosen to ensure a stable capture,especially for the non-cooperative target capture mission using multi-armed spacecraft.In this work scenario,the contact points on the base and on the arms are distributed on the opposite side of the target.Otherwise,large forces will be needed.To cope with this problem,an uneven-oriented distribution union criterion is proposed.The union criterion contains a virtual symmetrical criterion and a geometry criterion.The virtual symmetrical contact point criterion is derived from the proof of the force closure principle using computational geometry to ensure a stable grasp,and the geometry criterion is calculated by the volume of the minimum polyhedron formed by the contact points to get a wide-range distribution.To further accelerate the optimization rate and enhance the global search ability,a line array modeling method and a continuous-discrete global search algorithm are proposed.The line array modeling method reduces the workload of calculating the descent direction and the gradient available,while the continuous-discrete global search algorithm reducing the optimization dimension.Then a highly efficient grasping is achieved and the corresponding contact point is calculated.Finally,an exhaustive verification is conducted to numerically analyze the disturbance resistance ability,and simulation results demonstrate the effectiveness of the proposed algorithms.展开更多
For the remote sensing satellite with an unbalanced rotating payload suspended by an Active Magnetic Bearing(AMB),this paper proposes a payload-oriented control scheme where the high-precision payload attitude control...For the remote sensing satellite with an unbalanced rotating payload suspended by an Active Magnetic Bearing(AMB),this paper proposes a payload-oriented control scheme where the high-precision payload attitude control is dominating.Firstly,to suppress the disturbances induced by payload inertia uncertainties and state measurement errors,an integrated framework of parameter identification and nonlinear predictive filtering is proposed to estimate payload inertia parameters and system states from multi-timescale,noise-and drift-contaminated measurement data,breaking the mutual constraint between identification and filter.Secondly,based on the estimation results,the control law and bearing electromagnetic force allocation strategy of the payloadoriented scheme are provided,so that the payload tracks the desired motion and the satellite platform follows payload to prevent the air gap of AMB from exceeding the safety threshold.Finally,the simulations are carried out to verify the advantages of the proposed control scheme in enhancing the payload control precision and isolating the platform vibration.展开更多
Attitude planning of rigid bodies has many applications in robotics and aerospace.However,because the attitude configuration space is non-Euclidean and the constraints are complex and non-linear,the design of the atti...Attitude planning of rigid bodies has many applications in robotics and aerospace.However,because the attitude configuration space is non-Euclidean and the constraints are complex and non-linear,the design of the attitude curve has always been a tricky problem.In this paper,a gradient-based attitude planning method is proposed to simultaneously handle attitude pointing,angular velocity,torque,and time constraints on Lie group SO(3).Firstly,the attitude interpolation algorithm on SO(3)gives an attitude curve connecting the initial and target attitudes.The shape of the curve is determined by the fitting coefficients and maneuvering time.Secondly,to match the curve with suitable angular velocity and control torque,a nonlinear planning model with fitting coefficients and maneuver time as decision variables is proposed.Solving the problem gives a smooth attitude curve that satisfies both kinematic and dynamic constraints and also avoids complicated time allocation.Then,to apply the gradientbased solver,analytical formulas for the derivatives of each order of the attitude curve with respect to the decision variables are given in this paper.Finally,the effectiveness of the proposed algorithm is verified by a series of numerical simulations.展开更多
To meet the growing demand for autonomous assembly in unstructured environments,such as field exploration,disaster recovery,and space assembly,this study proposes a fixture-free assembly system requiring a single robo...To meet the growing demand for autonomous assembly in unstructured environments,such as field exploration,disaster recovery,and space assembly,this study proposes a fixture-free assembly system requiring a single robotic arm.We present a dual-layer framework comprising two components:An upper layer assembly sequence planner that generates feasible optimal assembly sequences through a branch-and-bound guided search algorithm,which incorporates geometric constraints,gravitational stability analysis,and load support capacity;a lower layer assembly manipulation planner that selects appropriate skills from predefined skill libraries while ensuring motion feasibility through real-time trajectory planning.In particular,we develop a novel skill organization mechanism that comprehensively considers motion constraints of multiple skill primitives during parameter configuration.Experimental validation in non-standardized environments demonstrates the robustness and effectiveness in managing complex,long-horizon assembly tasks,achieving notable improvements in efficiency and adaptability.展开更多
This paper solves the problem of model-free dual-arm space robot maneuvering after non-cooperative target capture under high control quality requirements.The explicit system model is unavailable,and the maneuvering mi...This paper solves the problem of model-free dual-arm space robot maneuvering after non-cooperative target capture under high control quality requirements.The explicit system model is unavailable,and the maneuvering mission is disturbed by the measurement noise and the target adversarial behavior.To address these problems,a model-free Combined Adaptive-length Datadriven Predictive Controller(CADPC)is proposed.It consists of a separated subsystem identification method and a combined predictive control strategy.The subsystem identification method is composed of an adaptive data length,thereby reducing sensitivity to undetermined measurement noises and disturbances.Based on the subsystem identification,the combined predictive controller is established,reducing calculating resource.The stability of the CADPC is rigorously proven using the Input-to-State Stable(ISS)theorem and the small-gain theorem.Simulations demonstrate that CADPC effectively handles the model-free space robot post operation in the presence of significant disturbances,state measurement noise,and control input errors.It achieves improved steady-state accuracy,reduced steady-state control consumption,and minimized control input chattering.展开更多
This study investigates the optimization configuration problem of control moment gyroscopes for flexible vibration suppression of large space structures.First,the structural dynamic model of an unconstrained plate-lik...This study investigates the optimization configuration problem of control moment gyroscopes for flexible vibration suppression of large space structures.First,the structural dynamic model of an unconstrained plate-like flexible spacecraft with control moment gyroscopes is established using the Lagrangian method and the finite element method.Second,an optimization problem is established with the positions of control moment gyroscope array as variables and the linear quadratic index of mechanical energy during vibration process as the objective function.Additionally,an improved gray wolf optimizer is used to solve the optimization problem.The optimization results show that the control moment gyroscopes should be primarily placed on the corners of the spacecraft in a symmetrical manner.Finally,the priority of control moment gyroscope installation positions was studied using topological iteration.The results indicate that the control moment gyroscopes should be primarily installed on the corners,followed by the central positions,and then gradually expanded outward.展开更多
As the constellation scale expands,the traditional constellation management mode imposes a substantial burden on ground stations.In order to construct a high-efficiency management mode for the low earth orbit(LEO)mega...As the constellation scale expands,the traditional constellation management mode imposes a substantial burden on ground stations.In order to construct a high-efficiency management mode for the low earth orbit(LEO)mega-constellation and to respond to the mission rapidly,a management strategy using distributed management domains as well as their dynamic evolution and maintenance methodology is proposed.In this paper,the distributed management domain is described as a variable topology consisting of groups of categorized satellites.The mega-constellation management topology is divided into a limited number of sub-topologies,determined by minimizing the average transmission latency and the frequency of management updates.Considering the dynamic of constellation,a method for predicting satellite management switching time is proposed,and a fast management maintenance strategy is designed to reassign satellites into new sub-topologies,ensuring a low overall update frequency of the management domain structure.Simulation validates that the management strategy divides the mega-constellation into dozens of management sub-topologies with similar structure and low-frequency management updates.Throughout the management period,each satellite remains under management with low transmission latency,and the overall management topology maintains long-term stability.展开更多
基金co-supported by the National Natural Science Foundation of China(No.12372045)the Guangdong Basic and Applied Basic Research Foundation,China(No.2023B1515120018)the Shenzhen Science and Technology Program,China(No.JCYJ20220818102207015).
文摘The increasing complexity of on-orbit tasks imposes great demands on the flexible operation of space robotic arms, prompting the development of space robots from single-arm manipulation to multi-arm collaboration. In this paper, a combined approach of Learning from Demonstration (LfD) and Reinforcement Learning (RL) is proposed for space multi-arm collaborative skill learning. The combination effectively resolves the trade-off between learning efficiency and feasible solution in LfD, as well as the time-consuming pursuit of the optimal solution in RL. With the prior knowledge of LfD, space robotic arms can achieve efficient guided learning in high-dimensional state-action space. Specifically, an LfD approach with Probabilistic Movement Primitives (ProMP) is firstly utilized to encode and reproduce the demonstration actions, generating a distribution as the initialization of policy. Then in the RL stage, a Relative Entropy Policy Search (REPS) algorithm modified in continuous state-action space is employed for further policy improvement. More importantly, the learned behaviors can maintain and reflect the characteristics of demonstrations. In addition, a series of supplementary policy search mechanisms are designed to accelerate the exploration process. The effectiveness of the proposed method has been verified both theoretically and experimentally. Moreover, comparisons with state-of-the-art methods have confirmed the outperformance of the approach.
基金supported by the National Natural Science Foundation of China(Nos.62003115,11972130)Shenzhen Natural Science Fund(the Stable Support Plan Program GXWD20201230155427003-20200821170719001).
文摘The contact point configuration should be carefully chosen to ensure a stable capture,especially for the non-cooperative target capture mission using multi-armed spacecraft.In this work scenario,the contact points on the base and on the arms are distributed on the opposite side of the target.Otherwise,large forces will be needed.To cope with this problem,an uneven-oriented distribution union criterion is proposed.The union criterion contains a virtual symmetrical criterion and a geometry criterion.The virtual symmetrical contact point criterion is derived from the proof of the force closure principle using computational geometry to ensure a stable grasp,and the geometry criterion is calculated by the volume of the minimum polyhedron formed by the contact points to get a wide-range distribution.To further accelerate the optimization rate and enhance the global search ability,a line array modeling method and a continuous-discrete global search algorithm are proposed.The line array modeling method reduces the workload of calculating the descent direction and the gradient available,while the continuous-discrete global search algorithm reducing the optimization dimension.Then a highly efficient grasping is achieved and the corresponding contact point is calculated.Finally,an exhaustive verification is conducted to numerically analyze the disturbance resistance ability,and simulation results demonstrate the effectiveness of the proposed algorithms.
基金the Open Fund of Heilongjiang Postdoctoral Fund,China(No.LBH-Z21141).
文摘For the remote sensing satellite with an unbalanced rotating payload suspended by an Active Magnetic Bearing(AMB),this paper proposes a payload-oriented control scheme where the high-precision payload attitude control is dominating.Firstly,to suppress the disturbances induced by payload inertia uncertainties and state measurement errors,an integrated framework of parameter identification and nonlinear predictive filtering is proposed to estimate payload inertia parameters and system states from multi-timescale,noise-and drift-contaminated measurement data,breaking the mutual constraint between identification and filter.Secondly,based on the estimation results,the control law and bearing electromagnetic force allocation strategy of the payloadoriented scheme are provided,so that the payload tracks the desired motion and the satellite platform follows payload to prevent the air gap of AMB from exceeding the safety threshold.Finally,the simulations are carried out to verify the advantages of the proposed control scheme in enhancing the payload control precision and isolating the platform vibration.
基金supported by Shenzhen Science and Technology Program(Grant No.JCYJ20220818102207015)Guangdong Basic and Applied Basic Research Foundation(Grant No.2023B1515120018).
文摘Attitude planning of rigid bodies has many applications in robotics and aerospace.However,because the attitude configuration space is non-Euclidean and the constraints are complex and non-linear,the design of the attitude curve has always been a tricky problem.In this paper,a gradient-based attitude planning method is proposed to simultaneously handle attitude pointing,angular velocity,torque,and time constraints on Lie group SO(3).Firstly,the attitude interpolation algorithm on SO(3)gives an attitude curve connecting the initial and target attitudes.The shape of the curve is determined by the fitting coefficients and maneuvering time.Secondly,to match the curve with suitable angular velocity and control torque,a nonlinear planning model with fitting coefficients and maneuver time as decision variables is proposed.Solving the problem gives a smooth attitude curve that satisfies both kinematic and dynamic constraints and also avoids complicated time allocation.Then,to apply the gradientbased solver,analytical formulas for the derivatives of each order of the attitude curve with respect to the decision variables are given in this paper.Finally,the effectiveness of the proposed algorithm is verified by a series of numerical simulations.
基金supported by the National Natural Science Foundation of China(12372045)the Guangdong Basic and Applied Basic Research Foundation(2023B1515120018)the Shenzhen Science and Technology Program(JCYJ20220818102207015)。
文摘To meet the growing demand for autonomous assembly in unstructured environments,such as field exploration,disaster recovery,and space assembly,this study proposes a fixture-free assembly system requiring a single robotic arm.We present a dual-layer framework comprising two components:An upper layer assembly sequence planner that generates feasible optimal assembly sequences through a branch-and-bound guided search algorithm,which incorporates geometric constraints,gravitational stability analysis,and load support capacity;a lower layer assembly manipulation planner that selects appropriate skills from predefined skill libraries while ensuring motion feasibility through real-time trajectory planning.In particular,we develop a novel skill organization mechanism that comprehensively considers motion constraints of multiple skill primitives during parameter configuration.Experimental validation in non-standardized environments demonstrates the robustness and effectiveness in managing complex,long-horizon assembly tasks,achieving notable improvements in efficiency and adaptability.
基金supported by the National Natural Science Foundation of China(No.12372045)the National Key Research and the Development Program of China(Nos.2023YFC2205900,2023YFC2205901)。
文摘This paper solves the problem of model-free dual-arm space robot maneuvering after non-cooperative target capture under high control quality requirements.The explicit system model is unavailable,and the maneuvering mission is disturbed by the measurement noise and the target adversarial behavior.To address these problems,a model-free Combined Adaptive-length Datadriven Predictive Controller(CADPC)is proposed.It consists of a separated subsystem identification method and a combined predictive control strategy.The subsystem identification method is composed of an adaptive data length,thereby reducing sensitivity to undetermined measurement noises and disturbances.Based on the subsystem identification,the combined predictive controller is established,reducing calculating resource.The stability of the CADPC is rigorously proven using the Input-to-State Stable(ISS)theorem and the small-gain theorem.Simulations demonstrate that CADPC effectively handles the model-free space robot post operation in the presence of significant disturbances,state measurement noise,and control input errors.It achieves improved steady-state accuracy,reduced steady-state control consumption,and minimized control input chattering.
基金supported by the Shenzhen Science and Technology Program(grant number JCYJ 20220818102207015).
文摘This study investigates the optimization configuration problem of control moment gyroscopes for flexible vibration suppression of large space structures.First,the structural dynamic model of an unconstrained plate-like flexible spacecraft with control moment gyroscopes is established using the Lagrangian method and the finite element method.Second,an optimization problem is established with the positions of control moment gyroscope array as variables and the linear quadratic index of mechanical energy during vibration process as the objective function.Additionally,an improved gray wolf optimizer is used to solve the optimization problem.The optimization results show that the control moment gyroscopes should be primarily placed on the corners of the spacecraft in a symmetrical manner.Finally,the priority of control moment gyroscope installation positions was studied using topological iteration.The results indicate that the control moment gyroscopes should be primarily installed on the corners,followed by the central positions,and then gradually expanded outward.
基金funded by State Key Laboratory of Micro-Spacecraft Rapid Design and Intelligent Cluster(MS02240109)Guangdong Basic and Applied Basic Research Foundation(2023B1515120018)Shenzhen Science and Technology Program(JCYJ20220818102207015)。
文摘As the constellation scale expands,the traditional constellation management mode imposes a substantial burden on ground stations.In order to construct a high-efficiency management mode for the low earth orbit(LEO)mega-constellation and to respond to the mission rapidly,a management strategy using distributed management domains as well as their dynamic evolution and maintenance methodology is proposed.In this paper,the distributed management domain is described as a variable topology consisting of groups of categorized satellites.The mega-constellation management topology is divided into a limited number of sub-topologies,determined by minimizing the average transmission latency and the frequency of management updates.Considering the dynamic of constellation,a method for predicting satellite management switching time is proposed,and a fast management maintenance strategy is designed to reassign satellites into new sub-topologies,ensuring a low overall update frequency of the management domain structure.Simulation validates that the management strategy divides the mega-constellation into dozens of management sub-topologies with similar structure and low-frequency management updates.Throughout the management period,each satellite remains under management with low transmission latency,and the overall management topology maintains long-term stability.
