A method of monocular vision relative pose based on Legendre orthogonal basis neural network is proposed. The BP neural network

RBF neural network and Legendre orthogonal basis neural network are combined to solve the strong nonlinear solution in relative position and attitude solution algorithm for monocular vision. The optimal weights in Legendre orthogonal basis neural network are derived from gradient descent method. The developed method can deal with the problem of strong nonlinearity in relative position and attitude solution algorithm for monocular vision. The computer simulations and semi-physical simulations show that the proposed method can obtain satisfactory accuracy

robustness and real-time.