This paper proposes a navigation system for mobile robot working in indoor environment, which includes mapping and trajectory generation for the motion tracking problem of the real mobile robot. The mapping process comprises a Gaussian model to improve the accuracy of the environment representation. With this new mapping method, the mobile robot knows the obstacle in the environment around robot itself. For the trajectory generation process, a multimodality method is used to explore all feasible paths to reach the target point by estimating the density of distribution of the objective function. Finally, the experiments show the effectiveness of this navigation framework.