Multi-sensor localization and navigation for remote manipulation in smoky areas

Abstract

When localizing mobile sensors and actuators in indoor environments laser meters, ultrasonic meters or even image processing techniques are usually used. On the other hand, in smoky conditions, due to a fire or building collapse, once the smoke or dust density grows, optical methods are not efficient anymore. In these scenarios other type of sensors must be used, such as sonar, radar or radiofrequency signals. Indoor localization in low-visibility conditions due to smoke is one of the EU GUARDIANS [1] project goals.

The developed method aims to position a robot in front of doors, fire extinguishers and other points of interest with enough accuracy to allow a human operator to manipulate the robot's arm in order to actuate over the element. In coarse-grain localization, a fingerprinting technique based on ZigBee and WiFi signals is used, allowing the robot to navigate inside the building in order to get near the point of interest that requires manipulation. In fine-grained localization a remotely controlled programmable high intensity LED panel is used, which acts as a reference to the system in smoky conditions. Then, smoke detection and visual fine-grained localization are used to position the robot with precisely in the manipulation point (e.g., doors, valves, etc.).