Date of Original Version
The final publication is available at Springer via http://dx.doi.org/10.1007/s13748-011-0007-1
Abstract or Description
We consider a set of static towers with communication capabilities, but not within range of each other, i.e., sparsely positioned in an environment with obstacles that degrade the communication signal, e.g., emergency teams in areas where connectivity has been lost. We address the problem of deploying mobile robots, initially not necessarily within range of each other or of the static towers, to be communication gateways among the towers. The robots do not know the environment, the tower positions, nor their own initial position in global coordinates. After connectivity is achieved, we use this team of robots to locate and tether to an autonomous agent, using only measurements of signal strengths, without the need for the agent to communicate with the team of robots. We first discuss the challenges of the domain. We then contribute our distributed algorithm, where robots share connectivity information without merging maps, acquire information through their navigation, and heuristically plan their exploration. The robots analyze their own accumulated knowledge and determine if a positioning plan exists to achieve the joint connectivity goal. We introduce different navigation heuristics for achieving the connectivity goal. We formally define the multi-robot tethering problem, and contribute our 2-step algorithm to solve this problem using a data-driven RSSI-distance model. We illustrate our connectivity algorithm in simulation and compare the efficiency of the proposed heuristics. We apply the most promising heuristic in a variety of realistic indoor scenarios, demonstrating its efficacy. We then perform experiments of our tethering algorithm in simulation and with real robots in an actual office environment and show that we successfully solve the tethering problem.
Progress in Artificial Intelligence, 1, 1, 3-23.