Classifying Obstacles and Exploiting Knowledge About Classes for Efficient Humanoid Navigation

Abstract

In this paper, we propose a new approach to humanoid navigation through cluttered environments that exploits knowledge about different obstacle classes and selects appropriate robot actions. To classify objects from RGB images and decide whether an obstacle can be overcome by the robot with a corresponding action, e.g., by pushing or carrying it aside or stepping over or onto it, we train a convolutional neural network (CNN). Based on the associated action costs, we compute a cost grid of the environment on which a 2D path can be efficiently planned. This path encodes the necessary actions that need to be carried out to reach the goal. We implemented our framework in ROS and tested it in various scenarios with a Nao robot. As the experiments demonstrate, using the CNN the robot can robustly classify the observed obstacles into the different classes and exploit this information to efficiently compute solution paths. Our system finds paths also through regions where traditional planning methods are not able to calculate a solution or require substantially more time.