Efficiency of the Visual FastSLAM Technique with a Common Feature Map for Two Vehicles in the Integrated Exploration of an Indoor Environment

[摘要] This study investigates the integrated exploration of an environment with two vehicles using an exploration strategy based on occupancy grid and a technique of simultaneous localization and mapping (SLAM) for location of the vehicles. The exploration uses concepts of cost and utility of frontier-cells. Furthermore, the used SLAM method is based on a FastSLAM algorithm with landmarks extracted of the environment from visual sensors using the scale-invariant feature transform technique and on a feature map common to all vehicles. Both activities—location of the vehicles and environment exploration—are coordinated by a central agent. The goals of this article are (a) to show, through experimental results obtained in simulated environments, that when two vehicles communicate with a central agent building a feature map common to vehicles, the exploration task becomes more efficient than that performed with dedicated maps, because the accuracy of position and orientation of the vehicles are increased and (b) present and evaluate the implementation of the approach by using an autonomous vehicle and sensors (stereoscopic visual and laser). The motivation for using a feature map common to all vehicles is that a vehicle does not need to return to a previously explored area in order to reduce the uncertainty of its pose. It can reduce the uncertainty when observes a landmark previously mapped by another vehicle...

[学科分类] 自动化工程

[关键词] SLAM ;FastSLAM ;Multiple vehicles ;Integrated exploration ;Motion control