UBC Agrobot’s Navigation sub-team develops the self-driving software for the Agrobot to autonomously navigate crop rows using an on-board camera. In our first year, our team explored methods of crop-row detection and developed software to facilitate real-time motion planning. We have worked together to research different methods of computer vision and error control algorithms and applied these concepts to Python code using OpenCV. The software loads live-stream video, employs a series of filtering and transformations to detect crop-rows, and then applies a centering algorithm to determine the direction of the AgroBot.
In our upcoming year, our team plans to improve the accuracy of our navigation software while also integrating the software with the motor control system. This entails incorporating additional computer vision techniques to our code, implementing a robust software controller for continuous error adjustment and incorporating ROS for communication between subsystems.
The Navigation team is looking for members interested in software design, real-time computer vision, robotics, and embedded systems to develop Agrobot’s self-drive technology!