Development of navigation, localization and mapping capabilities on MAVEN

Abstract

MAVEN is a robot with mecanum wheels which integrates telepresence, media and advanced robotic technologies. In order to achieve telepresence, it must possess the ability of basic navigation, localization and mapping. This project aims to develop the navigation, localization, and mapping functions on MAVEN.

MAVEN is not differential drive but holonomic drive. A specific kinematic model is needed to be developed in order to navigate the robot in holonomic drive mode. With the navigation ability, MAVEN will then be able to achieve advanced functions. In order to achieve waypoint navigation, the robot itself must be able to avoid obstacles, construct a map, localize itself in the map, and create a path from initial position to goal position.

In order to achieve basic navigation, GALIL motion controller is adopted to communicate in between the computer onboard robot and motors. All the navigation commands are executed through GALIL motion controller. By applying CARMEN system, MAVEN is able to construct a map, localize itself, avoid obstacles and do path planning. However, all these functions are operated in differential drive mode. Therefore, another system named ‘’ROS’’ is adopted in order to accomplish all these tasks in mecanum drive mode.

Experimental result showed that the MAVEN can navigate in mecanum drive mode precisely through the kinematic model developed (only in a short period of drive). Also, the mapping and localization performances are considered satisfactory. However, the odometry error accumulates for long run and it will affect the performances of navigation, localization and mapping functions.