SLAM algorithms implementation in a UAV, based on a heterogeneous system: A survey

摘要

Simultaneous localization and mapping (SLAM) is a solid problem in robotics, this problem can be solved using an algorithm dedicated to the resolution of this problem. In this context we find large variety algorithms like FastSLAM2.0, GraphSLAM, EKFSLAM, VoSLAM, ORBSLAM and other. The drone by its ease and speed to implement allows realizing various types of aerial mapping increasingly used in many sectors of professional activities such as building, public works or agriculture. In this study we propose an implementation of the VoSLAM algorithm in a UAV, based on a heterogeneous system using the OpenCL parallel programming. In the agricultural cities, for agricole process parameters controlling, we can use the heterogeneous DE1-Soc architecture of VoSLAM algorithm based on the OpenCL parallel programming. This architecture permits the control in real-time of the humidity based on a treatment of image to transmit thereafter its data..