Our first paper has been accepted at the IEEE/ASME international conference on Advanced Intelligent Mechatronics (AIM). It will take place in Sapporo, Japan next July. The article focuses on the control of the mobile robot and presents our first results on this topic which belongs to WP1. The proposed strategy uses the visual data provided by the embedded Real Sense cameras and relies on a sensor-based model predictive control law to autonomously navigate through the orchard. Its originality is that the designed control law is coupled with a spiral-based framework allowing to deal similarly with the two main tasks involved in the movement: the in-row navigation and headland maneuvers. The Model-based Predictive Control scheme allows to deal with several important constraints such as stability, actuator boundaries and image plane limits. It is also taken advantage of the VPC principle to enable a switch between the two previously mentioned tasks. This work has been validated using our orchard simulator based on ROS and Gazebo. We are currently implementing this strategy on our Hunter 2 robot and plan experimental tests during summer in orchards. The video proposed hereafter shows the obtained results and demonstrates the interest of the approach.
All scientific details can be found in this reference: Autonomous Navigation Strategy for Orchards Relying on Sensor-based Nonlinear Model Predictive Control, A. Villemazet, A. Durand-Petiteville, V. Cadenat. IEEE/ASME International Conference on Advanced Intelligent Mechatronics (AIM), Sapporo, Japan, July 11th – 15th 2022.