MANTRA Autonomous Agricultural Services Vehicle

Sean Finnigan

Abstract


Intelligent vehicle solutions are getting a lot of press lately, with autonomous vehicle technology being at the forefront of many newsworthy tech stories. Many of these stories focus on the impending flood of expensive consumer products soon entering the market. Until recently, machine vision based autonomy was out of reach for most people/organizations interested in simultaneous localization and mapping (SLAM) problems. Lately however, sensor affordability and open source software resources have made accessibility a reality. In this paper, we provide a simple, affordable technique for solving a mobile robot SLAM problem. Our research shows that by using open source software, companies and individuals alike can access these technologies without breaking the bank. Our team created an autonomous agricultural services vehicle for ARGO ATV, a company that manufactures amphibious electric vehicles.  In exchange for a donated chassis, ARGO asked us to design a system that can safely navigate and map a banana orchard (for applications in Martinique) without direct user input. We achieved this by leveraging the resources of the Robot Operating System (ROS) and through collaboration with the open source robotics community. Our system integrates information from a high resolution stereo vision camera, 2D LIDAR, Inertial Measurement Unit, and a GPS module to implement a real-time SLAM (Simultaneous Localization and Mapping) algorithm on a high performance, low power Embedded System. Our vehicle can navigate autonomously while safely avoiding obstacles, animals, and farm workers.

Keywords


SLAM, ROS, Autonomous Navigation

Full Text: PDF

Refbacks

  • There are currently no refbacks.