论文部分内容阅读
End to end motion control is a new paradigm for mobile robots,it maps sensors output to motion control commands directly.In this paper,an approach to end to end control based on a 2D laser range finder is proposed.And a convolutional neural network and a fully connected network are constructed for controlling velocity and angular rate respectively.With an existing simulation system,the training dataset and test dataset are generated.In a simulation environment,the proposed method is tested and verified with the generated test datasets.The test experiments show that the method is feasible,but it is need to be improved in several aspects.