Smoothing-homotopy-based sequential convex programming for trajectory optimization

This paper proposes a new smoothing-homotopy-based sequential convex programming (SCP) method for general trajectory optimization problems. The surrogates, derived from convolving the smoothing kernel with the terminal states, are firstly incorporated into the terminal constraints as replacements of...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:Aerospace science and technology Ročník 159; s. 109995
Hlavní autoři: Zhao, Mengxin, Pan, Binfeng, Hou, Xiyu, Huang, Longxin
Médium: Journal Article
Jazyk:angličtina
Vydáno: Elsevier Masson SAS 01.04.2025
Témata:
Modified Chebyshev–Picard iteration
Sequential convex programming
Smoothing homotopy
Trajectory optimization
Smoothing homotopy
Sequential convex programming
Modified Chebyshev–Picard iteration
Trajectory optimization
ISSN:1270-9638
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:This paper proposes a new smoothing-homotopy-based sequential convex programming (SCP) method for general trajectory optimization problems. The surrogates, derived from convolving the smoothing kernel with the terminal states, are firstly incorporated into the terminal constraints as replacements of the original ones. The smoothing parameter, taken as the homotopic parameter, decreases from a larger value to zero, with which the corresponding optimization problem gradually transitions from an easier and smoothed counterpart to the original one. Both the modified Chebyshev-Picard iteration (MCPI) approach and the trapezoidal rule are employed to transcribe the continuous-time optimization problem into a series of finite-dimensional sub-problems, respectively, which are then solved by the primal-dual interior-point solver with the aid of convexity techniques. Numerical simulations for an ascent trajectory optimization problem are provided to demonstrate the performance of the proposed method, showcasing its superior convergence compared to the standard SCP methods. •A novel smoothing-homotopy-based sequential convex programming (SCP) method is proposed.•The smoothing kernel convolves the terminal states within the time domain, replacing the original states in the SCP problem.•The modified Chebyshev-Picard iteration approach is employed.•The convergence and efficiency of the proposed method are validated through Monte Carlo tests.
ISSN:1270-9638
DOI:10.1016/j.ast.2025.109995