Trajectory Prediction of a Flying Object Based on Hybrid Mapping Between Robot and Camera Space
Tao Xue, Yong kang Liu · 2018
Trajectory prediction of flying objects holds important significance for research and utilization in military, industrial and other fields. This paper presents a method for the trajectory prediction of a flying object in camera space. We propose a method combining global and local mapping model to calculate where in the world a specific camera space object is. With global mapping we can ensure accuracy while exploring the optimal size of data set and the updating method of training data. When the object is similar to our training data, we apply the local model and use fewer training sets to obtain the mapping. After that, the two-dimensional trajectory in the image is transformed into the three-dimensional trajectory in camera space based on different spatial mapping parameters. We use the SVR algorithm to predict the trajectory of a flying object. Experiments show that the mapping model we proposed can be well combined with the SVR method to get a more accurate prediction trajectory. During the flight, we constantly modify the predicted trajectory, then we can get the more accurate grasp point of the manipulator, so as to achieve the grasp for flying objects by a manipulator.