Computer Vision System
One of the key task for the AgriRobot system develop is understanding the agricultural environment for both the navigation and spraying systems. To enable this visual understanding of the landscape RGB-Depth cameras where deployed on the robot to capture the visual data and computer vision and AI methods used for detecting objects within an image. The system developed by the researchers at Loughborough applied deep learning based semantic segmentation with depth data fusion to provide accurate object information while also running at real-time on the target hardware of an Nvidia Jetson TX2 device.
Overview of the computer vision system
Semantic Segmentation Architecture