arXiv subjects

Fanghua Jiang

Publications and source records attributed to Fanghua Jiang.

9 recordsLinked to original sources

arXiv2026

Flow-Matched Motion Priors: Online Optimal-Transport Rewards for Imitation Learning

Learning a motion prior requires a reward that guides a policy from its current behavior toward demonstrated motion. Adversarial Motion Priors (AMP) provide such a reward with a discriminator. However, adversarial objectives can become uninformative when policy and expert supports are far apart. A naive use of optimal transport (OT) averages matched expert successors into a barycentric target. Averaging across gait phases can weaken the target's joint motion. We introduce Flow-Matched Motion Priors (FMP), an online scalar reward learned from paths connecting current rollout histories to an expert motion bank. Entropic OT supplies the coupling. Before each policy update, we train a neural potential with flow matching (FM) along the rollout-to-expert paths, endpoint-gradient supervision, and relative-value calibration. The actor receives only physical observations and the reward remains a scalar, as in AMP. Controlled reward-model experiments show substantially better generalization beyond the fitting rollout than value-only or endpoint-only fitting. On Unitree G1, matched 50-million-transition experiments compare FMP with AMP, a barycentric OT reward, and nested ablations under demonstration and fixed-pose initialization. FMP produces stable forward walking at 0.727 m/s from demonstration resets and 0.338 m/s from a fixed default pose. In the fixed-pose condition, it incurs 129 falls versus 243 for the endpoint-only control. Against a static score-gradient teacher, dynamic FM reduces score-increment error at interpolation fractions 0.25 and 0.50 while using 29% less offline fitting time.

cs.RO↗

arXiv2026

Parallel-in-Time Nonlinear Optimal Control via GPU-native Sequential Convex Programming

Real-time solution of nonlinear optimal control problems remains challenging on embedded robotic hardware, where conventional solvers often rely on global sparse linear algebra or sequential recursions that are difficult to map efficiently to massively parallel processors. This paper presents ucenter, a GPU-native Sequential Convex Programming (SCP) framework for nonlinear optimal control. At each SCP iteration, nonlinear dynamics are linearized around a nominal trajectory, and the resulting convexified subproblem is solved by a consensus Alternating Direction Method of Multipliers (ADMM) scheme. The temporal splitting replaces global sparse Karush-Kuhn-Tucker factorizations with independent per-node dense solves, closed-form dynamic consistency updates, and analytical projections onto convex constraint sets. Both the outer SCP loop and the inner ADMM subproblem are executed entirely on the GPU, enabling efficient optimization. The proposed solver is evaluated on quadrotor obstacle avoidance and Mars powered descent problems using an NVIDIA Jetson AGX Orin edge platform. Benchmarking against a CPU-parallel iLQR baseline in randomized environments reveals that the GPU implementation achieves over 100 Hz batched planning throughput, a 4.1x speedup, and a 51% reduction in energy consumption, while consistently maintaining low nonlinear dynamics defects. The framework exposes reusable GPU-parallel optimization primitives that can be specialized to a wide variety of complex nonlinear optimal control settings, as demonstrated by the scenario-based robust MPC and batched Monte Carlo generation tasks.

cs.RO↗

arXiv2025

Multi-Revolution Low-Thrust Trajectory Optimization With Very Sparse Mesh Pseudospectral Method

Multi-revolution low-thrust trajectory optimization problems are important and challenging in space mission design. In this paper, an efficient, accurate, and widely applicable pseudospectral method is proposed to solve multi-revolution low-thrust trajectory optimization problems with various objective functions and perturbations. The method is based on the Sundman transformation and pseudospectral method, together with a sparse mesh that is monotonic, near-uniformly spaced, and uniformly scattered on the unit circle. Two methods are proposed to construct the mesh: a deterministic method based on rotation mapping; a stochastic method utilizing autocorrelated random sequences. Core mechanisms ensuring the correctness of the method are analyzed, including the dual roles of mesh points as both integration points in the temporal domain and sampling points in the angular domain, the slow dynamics of the system excluding the fast angle variable, and the nearly commutative vector fields generated by applying different control inputs. The method is demonstrated through a multi-revolution low-thrust orbital rendezvous problem. Results show that the proposed method achieves high accuracy with only a few seconds of computational time for challenging problems.

eess.SY↗

arXiv2025

Re-examining the Legendre-Gauss-Lobatto Pseudospectral Methods for Optimal Control

Pseudospectral methods represent an efficient approach for solving optimal control problems. While Legendre-Gauss-Lobatto (LGL) collocation points have traditionally been considered inferior to Legendre-Gauss (LG) and Legendre-Gauss-Radau (LGR) points in terms of convergence properties, this paper presents a rigorous re-examination of LGL-based methods. We introduce an augmented formulation that enhances the standard LGL collocation approach by incorporating an additional degree of freedom (DOF) into the interpolation structure. We demonstrate that this augmented formulation is mathematically equivalent to the integral formulation of the LGL collocation method. Through analytical derivation, we establish that the adjoint system in both the augmented differential and integral formulations corresponds to a Lobatto IIIB discontinuous collocation method for the costate vector, thereby resolving the previously reported convergence issues. Our comparative analysis of LG, LGR, and LGL collocation methods reveals significant advantages of the improved LGL approach in terms of discretized problem dimensionality and symplectic integration properties. Numerical examples validate our theoretical findings, demonstrating that the proposed LGL-based method achieves comparable accuracy to LG and LGR methods while offering superior computational performance for long-horizon optimal control problems due to the preservation of symplecticity.

eess.SY↗

arXiv2025

Vectorized Sparse Second-Order Forward Automatic Differentiation for Optimal Control Direct Methods

Direct collocation methods are widely used numerical techniques for solving optimal control problems. The discretization of continuous-time optimal control problems transforms them into large-scale nonlinear programming problems, which require efficient computation of first- and second-order derivatives. To achieve computational efficiency, these derivatives must be computed in sparse and vectorized form, exploiting the problem's inherent sparsity structure. This paper presents a vectorized sparse second-order forward automatic differentiation framework designed for direct collocation methods in optimal control. The method exploits the problem's sparse structure to efficiently compute derivatives across multiple mesh points. By incorporating both scalar and vector nodes within the expression graph, the approach enables effective parallelization and optimized memory access patterns while maintaining flexibility for complex problems. The methodology is demonstrated through application to a prototype optimal control problem. A complete implementation for multi-phase optimal control problems is available as an open-source package, supporting both theoretical research and practical applications.

eess.SY↗

arXiv2022

An Atlas of Optimal Low-Thrust Rephasing Solutions in Circular Orbit

In this paper, the time- and propellant-optimal low-thrust rephasing problems in circular orbit are studied to depict their solution spaces in an atlas. The number of key parameters that settle the rephasing problems is reduced by developing a set of linearized equations of motion based on the Sundman transformation and by formulating two reduced shooting functions using the minimum principle and symmetry properties. Only one key parameter is identified for the time-optimal problem, while two key parameters are obtained for the propellant-optimal one. Numerical investigation of the relationships between these parameters and shooting variables reveals that they can be depicted by some curve (or contour) maps and approximated by piecewise functions (or linear interpolations). For the relatively short- or long-term rephasing cases, some analytical time- and propellant-optimal solutions are proposed and consistent with the numerical solutions. Numerical results demonstrate that the proposed solutions can provide good initial guesses to solve the low-thrust rephasing problems with nonlinear dynamics. Moreover, the approximations of the performance indexes can be used in the preliminary mission design.

math.DS↗

arXiv2022

Analytical Shaping Method for Low-Thrust Rendezvous Trajectory Using Cubic Spline Functions

Preliminary mission design requires an efficient and accurate approximation to the low-thrust rendezvous trajectories, which might be generally three-dimensional and involve multiple revolutions. In this paper, a new shaping method using cubic spline functions is developed for the analytical approximation, which shows advantages in the optimality and computational efficiency. The rendezvous constraints on the boundary states and transfer time are all satisfied analytically, under the assumption that the boundary conditions and segment numbers of cubic spline functions are designated in advance. Two specific shapes are then formulated according to whether they have free optimization parameters. The shape without free parameters provides an efficient and robust estimation, while the other one allows a subsequent optimization for the satisfaction of additional constraints such as the constraint on the thrust magnitude. Applications of the proposed method in combination with the particle swarm optimization algorithm are discussed through two typical interplanetary rendezvous missions, that is, an inclined multi-revolution trajectory from the Earth to asteroid Dionysus and a multi-rendezvous trajectory of sample return. Simulation examples show that the proposed method is superior to existing methods in terms of providing good estimation for the global search and generating suitable initial guess for the subsequent trajectory optimization.

cs.RO↗

arXiv2019

Real-Time Optimal Control for Irregular Asteroid Landings Using Deep Neural Networks

Precise soft landings on asteroids are central to many deep space missions for surface exploration and resource exploitation. To improve the autonomy and intelligence of landing control, a real-time optimal control approach is proposed using deep neural networks (DNN) for asteroid landing problems wherein the developed DNN-based landing controller is capable of steering the lander to a preselected landing site with high robustness to initial conditions. First, to significantly reduce the time consumption of gravity calculation, DNNs are used to approximate the irregular gravitational field of the asteroid based on the samples from a polyhedral method. Then, an approximate indirect method is presented to solve the time-optimal landing problems with high computational efficiency by taking advantage of the designed gravity approximation method and a homotopy technique. Furthermore, five DNNs are developed to learn the functional relationship between the state and optimal actions obtained by the approximate indirect method, and the resulting DNNs can generate the optimal control instructions in real time because there is no longer need to solve the optimal landing problems onboard. Finally, a DNN-based landing controller composed of these five DNNs is devised to achieve the real-time optimal control for asteroid landings. Simulation results of the time-optimal landing for Eros are given to substantiate the effectiveness of these techniques and illustrate the real-time performance, control optimality, and robustness of the developed DNN-based optimal landing controller.

math.OC↗

arXiv2016

Systematic Low-Thrust Trajectory Optimization for a Multi-Rendezvous Mission using Adjoint Scaling

A deep-space exploration mission with low-thrust propulsion to rendezvous with multiple asteroids is investigated. Indirect methods, based on the optimal control theory, are implemented to optimize the fuel consumption. The application of indirect methods for optimizing low-thrust trajectories between two asteroids is briefly given. An effective method is proposed to provide initial guesses for transfers between close near-circular near-coplanar orbits. The conditions for optimality of a multi-asteroid rendezvous mission are determined. The intuitive method of splitting the trajectories into several legs that are solved sequentially is applied first. Then the results are patched together by a scaling method to provide a tentative guess for optimizing the whole trajectory. Numerical examples of optimizing three probe exploration sequences that contain a dozen asteroids each demonstrate the validity and efficiency of these methods.

astro-ph.IM↗