A Sensor Fusion Layer to Cope with Reduced Visibility in SLAM

Abstract

Mapping and navigating with mobile robots in scenarios with reduced visibility, e.g. due to smoke, dust, or fog, is still a big challenge nowadays. In spite of the tremendous advance on Simultaneous Localization and Mapping (SLAM) techniques for the past decade, most of current algorithms fail in those environments because they usually rely on optical sensors providing dense range data, e.g. laser range finders, stereo vision, LIDARs, RGB-D, etc., whose measurement process is highly disturbed by particles of smoke, dust, or steam. This article addresses the problem of performing SLAM under reduced visibility conditions by proposing a sensor fusion layer which takes advantage from complementary characteristics between a laser range finder (LRF) and an array of sonars. This sensor fusion layer is ultimately used with a state-of-the-art SLAM technique to be resilient in scenarios where visibility cannot be assumed at all times. Special attention is given to mapping using commercial off-the-shelf (COTS) sensors, namely arrays of sonars which, being usually available in robotic platforms, raise technical issues that were investigated in the course of this work. Two sensor fusion methods, a heuristic method and a fuzzy logic-based method, are presented and discussed, corresponding to different stages of the research work conducted. The experimental validation of both methods with two different mobile robot platforms in smoky indoor scenarios showed that they provide a robust solution, using only COTS sensors, for adequately coping with reduced visibility in the SLAM process, thus decreasing significantly its impact in the mapping and localization results obtained.

A Comprehensive Review of Applications of Drone Technology in the Mining Industry

“…Ultrasonic sensors are low-cost sensors for obstacle detection that have been tested by several studies. Other kinds of obstacle detector sensors that are repeatedly used by researchers include infrared sensors, stereo cameras, and laser range finders (LRFs) [75][76][77][78][79][80].…”

Keyframes Global Map Establishing Method for Robot Localization through Content-Based Image Matching

“…For this research field, SLAM (Simultaneous Localization and Mapping) is the commonly employed [3], especially the V-SLAM (Visual Simultaneous Localization and Mapping) [4,5]. Many latest variants take advantage of new features or 3D information to build navigation map [6,7], including ORB SLAM [8], dense SLAM [9,10], semidense SLAM [11], LSD SLAM [12], and CV-SLAM [13,14].…”

Fusing Range Measurements from Ultrasonic Beacons and a Laser Range Finder for Localization of a Mobile Robot

“…Aside from the filtering methods, adaptive estimation [ 29 ] and fuzzy logic-based methods [ 30 ] have been proposed. The adaptive estimation method [ 29 ] fuses vision, odometry and an attitude heading reference system for use in an indoor environment.…”

“…This method extracts velocity using vision to apply visual information in the adaptive estimation formulation. The fuzzy logic-based method [ 30 ] fuses LRF measurements with sonar array range sensor measurement. Unlike the ultrasonic beacons used in our research, the sonar array range sensor detects the range from a sonar transducer array attached on the robot to objects that block the sonar beam.…”