Grid Based Navigation and Robust Control of Multiple Networked Mobile Robots

Thumbnail Image
Issue Date
Kim, Doug
The Graduate School, Stony Brook University: Stony Brook, NY.
The field of robotics has been drawing much interest from academia and industry alike. However its applications have been limited to static models such as factory assembly lines where it performs repetitive actions that are precisely programmed. This is to minimize the human interaction and possibilities of any deviation from prescribed actions such as navigation errors in case of mobile robots. The robots being used in many different application also have very limited interactions between robots themselves. The collaboration between the robots can increase the efficiency of the job they are programmed to do as well as assisting in situations where a robot is in recoverable and non-recoverable failure mode. The collaboration between robots necessitates a reliable communication network as well as a robust control mechanism.In this paper, first, the grid based navigation is presented where the area robots assigned to navigate is divided into grids. The grid based navigation approach simplifies the navigation algorithm with a certain degree of tolerance. The size of grid depends on the resolution of data robots are collecting. A numerical analysis of an error tolerance for an application of RF path loss data collection is studied to show the effect of grid sizes. We also take into account a range of the communication network between robots in calculating the size of grids to ensure the reliability of the communication network. The navigation algorithm is based on using laser range finders for estimating and deciding the current grid locations that the robots are in. The path planning methods for robots are also presented with precise timing as required in a multi-robot collaboration. The script-based path planning reduces risk of robots interfere with each other while navigating to their target grids. The scripts are pre-generated based on the grid configuration and the number of robots being utilized.Secondly, A novel robust robot control algorithm is presented so that when there is any failure in robots, the robots can reconfigure themselves to complete the task on hand. A robust control algorithm is devised to prevent catastrophic failure at the system level as the algorithm can determine severity of the problem and take an action to isolate and mitigate the problems caused by failed robots. The robust control algorithm also aids the navigation where the navigation scripts are not executed properly by robots. Since the scripts are pre-generated, we use the robust control algorithm to reconfigure and regenerate scripts for the robots.Finally, the simulation and experimental results are presented using the grid based navigation and robust control algorithm. The simulation shows how the robots estimate the current grid location using boundary information from the rangefinder measurement data. The simulation and experiment data also show how the navigation method responds to different building configurations and properly estimate robots' locations.\indent The paper finds that the grid based navigation and robust control algorithm can be applied to multiple mobile robots to enable collaboration between them while minimizing the probability of system level catastrophic failures.