Abstract:This paper describes a cognitive map building and navigation system using an RGB-D sensor for mobile robots. A brain-inspired simultaneously localization and mapping (SLAM) system, that requires raw odometry data and RGB-D information, is used to construct a spatial cognitive map of an office environment. The cognitive map contains a set of spatial coordinates that the robot has traveled. A global path is extracted from the built cognitive map and subsequently used by a local planner to instruct the robot to navigate. The global path is a subset of the path that builds up the cognitive map. This is different from other path planning mechanisms that construct a path based on a ground-truth map. Experiment results show that the employment of the RGB-D sensor significantly improves the mapping results.

Rgb-D Based Cognitive Map Building And Navigation