Onboard trajectory generation for entry vehicles via adaptive multivariate pseudospectral interpolation

The multivariate pseudospectral interpolation approach has been coupled with an algorithm of subspace selection to be able to generate online nearly optimal real-time trajectories for entry scenarios in the presence of wide dispersions at the entry interface. The Monte Carlo campaign has demonstrated the feasibility of this approach, having as further advantage a significant improvement in the guidance performances, analyzed both in terms of longitudinal error and in terms of footprint dispersion, reduced to =6% of the one obtained with a more traditional NTT approach. A further advantage coming from the use of this technique is its lossless reduction of the database size. High-density discrete trajectories can be obtained by storing a significantly smaller number of values, reducing the size to =3% of a database stored in the traditional way, and this aspect directly relaxes the requirements for the memory of the onboard CPU.