VISION-BASED CONTROL IN DRIVING ASSISTANCE OF AGRICULTURAL VEHICLES

This article presents a real-time control system for an agricultural m obile machine (vehicle) based on an on-board vision system using a sin gle camera. This system has been designed to help humans in repetitive and difficult tasks in the agricultural domain. The aim of the roboti cs application concerns the control of the vehicle with regard to the reap limit detected in image space. The perception aspect in relation to the application has been described in previous work and here we dea l with the control aspect. We integrate image features issued from the modeling of the scene in the control loop to perform an image-based s ervoing technique. The vehicle behavior described here concerns bicycl e and neural models, and three control laws are then synthesized The f irst and the second are modeling approaches and use an interaction bet ween the scene and the image space. They are based on the regulation o f a task function The third is a ''black-box modeling'' technique, and is based on a neural network. Finally, experimental results obtained with these different control laws in different conditions are presente d and discussed.