In order to further improve the penetration ability of hypersonic vehicle, the low-observable glide trajectory optimization method for hypersonic vehicle was proposed. The radar cross section of 180°×360° direction of the aircraft was considered. The Gaussian filtering method was used to smoothing the original data, which was characterized by many spikes. This can improve the convergence of optimization problems without changing the trend of the original data. In order to enable the RCS (radar cross section) data to have strong fidelity, the cubic spline interpolation method was used to call the discrete data. The modelling of the low detectable glide trajectory optimization problem of hypersonic vehicle was completed. Setting the radar detection probability as the objective function, the hp adaptive Radau pseudo-spectral method was used to solve the problem. The stepwise calculate strategy was adopted to further improve the optimization efficiency and convergence performance. Simulation results show that when compared with the trajectory who takes the shortest flight time as the goal, the given method can effectively reduce the radar detection probability.