An Outdoor Navigation System Dedicated to a Moroccan Micro-tractor Based on SLAM Algorithms and Multi-sensor Fusion

Over the recent years, agricultural mobile robots have been an important topic for scientists and for the world of industry. The rapid progress of communication, sensor, and computing technologies has led to a significant increase in the field of guidance systems for autonomous agricultural robots. Agricultural robots that are automated decrease labor expenses, avoid farmers from performing different tasks, and give them reliable, up-to-date data to aid in management choices. This paper provides and discusses a description of the navigation mechanism for the "FellahBot" micro-tractor developed by FellahTech. Navigation sensors, computational methods, and navigation control algorithms are the essential components. Crucial operations include selecting, coordinating, and combining the most suitable sensors to provide the essential data needed for the robot's navigation. In order to achieve improved localization and mapping, image processing and multi-data sensor fusion employ powerful algorithms. Its lines of research are grouped under the name simultaneous localization and mapping (SLAM) algorithms. This scientific work aims to evaluate SLAM systems embedded on the architecture CPU-GPU of the Jetson Nano, and we compare with two implementations of CartographerSLAM and EKF algorithms.