Remote Low Frequency State Feedback Kinematic Motion Control for Mobile Robot Trajectory Tracking

Teleoperated robots generally receive high level commands from a remote system, while accomplishing motion control through conventional means. We present a teleoperated system that removes the entire motion control structure from the robot, in order to preserve the availability of crucial on- board resources. The operation of state feedback control is performed by a system remote from the robot. We have de- signed a computerized motion planning and control system for Mobile Emulab, and in this article, discuss the implementation of trajectory tracking control. A component of the Emulab network testbed, Mobile Emulab is used for wireless network experiments requiring mobility; and is publicly available to remote researchers via the Internet. Medium scale wheeled mobile robot couriers are used to move wireless antennas within a semi-controlled environment. Experimenters use a web-based GUI to specify desired paths and configurations for multiple robots. State feedback is provided by an overhead camera based visual localization system. Kinematic control is used to generate velocity commands, which are sent to robots over a computer network. Data availability is restricted to a low sampling frequency. There is significant noise, loss, and phase lag present in the robot localization data, which our research overcomes to provide an autonomous trajectory tracking mobile robot control system. Index Terms—Mobile robots, Motion control, Telerobotics I. INTRODUCTION This research addresses the engineering challenges en- countered while designing and implementing a system pro- viding teleoperation and remote feedback control of mobile robots over a network. The motion control system presented as part of this research is added as a component to Mobile Emulab (12), which is part of the Emulab network testbed (22). An overview of the Mobile Emulab system architecture is shown in Figure 1. The major components of Mobile Emulab are displayed, with their interactions denoted. The motion controller passes desired wheel velocities to the robots. The cameras pass image data to the localization system, which is used to determine the positions of all robots in the testbed environment. Robots are controlled over a wireless network and localized by a computer vision system which utilizes an overhead grid of video cameras. The motion controller presented in this paper is motivated by the need for smooth, precise motion control of robots