This paper presents a novel online trajectory optimization algorithm for hypersonic entry, specifically designed to address the challenges of advanced guidance applications. A key feature of the algorithm is its exploitation of the parallelizable nature of sequential convex programming and the interior-point method, which significantly enhances computational efficiency in both the outer and inner loops. The algorithm is also specifically optimized for embedded onboard processors, ensuring reliable performance in real-world applications. First, a parallel-improved first-order-hold discretization method is introduced, combined with an adaptive mesh update strategy, which improves the efficiency of outer-loop iterations while ensuring the dynamic feasibility of the solutions. Next, a customized parallel interior-point method solver is developed to efficiently distribute the computational load of solving the time-intensive Karush–Kuhn–Tucker systems across multicore processors. Additionally, a warm-start strategy is incorporated into the solver to accelerate convergence. Numerical experiments conducted on a surrogate onboard computer utilizing a multicore digital signal processor demonstrate that the proposed algorithm achieves high computational efficiency, with runtimes under 950 ms for typical multiconstraint entry scenarios, while maintaining high solution accuracy. This work represents a significant advancement in the development of high-performance, embedded trajectory optimization algorithms for autonomous aerospace guidance missions.
Wang et al. (Sun,) studied this question.