Conventional solutions to control design problems in complex space missions are often based on a solution of an associated optimal control problem. Due to limited onboard computing capacity, the computation of an optimal trajectory and a dedicated controller are commonly carried out offline, in advance of a mission. This approach requires the definition of a nominal set of initial conditions and system parameters, which are generally different from the conditions during the actual manoeuvre. This paper proposes an approach based on multivariate spline interpolation for generating nearly optimal solutions online, despite the difference between the nominal and actual conditions during the manoeuvre. The onboard computations are limited to simple interpolations and hence do not cause considerable additional computing effort. Compared to conventional control design approaches, the proposed method significantly increases the region of initial conditions and size of parametric uncertainties for which the manoeuvre can be successfully completed. The applicability and practical relevance of this approach are demonstrated for a Moon landing manoeuvre.