One path generation method of visual robot based on BLD

Robot learning by demonstration has become one of the most active research topics in robotics, and trajectory generation as one of the three points of demonstration study has become a hot research topic. Trajectory generation is decided to demonstrate learning an important factor in the success, the traditional method using sift tracking direct trajectory generation, but this method exists many shortcomings, such as more feature points selected tracks difficulties, trajectory there are certain noise, and so on. In this paper, a combination of SIFT, PCA and UKF algorithm algorithm based on trajectory generation method, the simulations and experiments with a real robot operation proves the validity of the algorithm.