End-to-end deep learning for autonomous navigation of mobile robot

摘要

This paper proposes an end-to-end method for training convolutional neural networks for autonomous navigation of a mobile robot. Traditional approach for robot navigation consists of three steps. The first step is extracting visual features from the scene using the camera input. The second step is to figure out the current position by using a classifier on the extracted visual features. The last step is making a rule for moving the direction manually or training a model to handle the direction. In contrast to the traditional multi-step method, the proposed visuo-motor navigation system can directly output the linear and angular velocities of the robot from an input image in a single step. The trained model gives wheel velocities for navigation as outputs in real-time making it possible to be implanted on mobile robots such as robotic vacuum cleaner. The experimental results show an average linear velocity error of 2.2 cm/s and average angular velocity error of 3.03 degree/s. The robot deployed with the proposed model can navigate in a real-world environment by only using the camera without relying on any other sensors such as LiDAR, Radar, IR, GPS, IMU.