Mark II: Design and Control of an Autonomous Mobile Agricultural Robot

Abstract

The project mark II with the theme Autonomous robot navigation through a corn field is concerned with an autonomous skid-steered robot. The background for the project is The Field Robot Event 2011, where robots compete in driving through a corn field with obstacles. There are clear rules regarding the competition, where only autonomous robots can compete. The goal of the competition is to drive through a corn field without knocking down any plants, being able to turn the robot at the end of a passage and detect and mark weeds scattered across the field. Detecting and marking weeds has not been part of this project. Through a selection process of available platforms, the platform, that fulfilled set up criteria the best, was chosen. The chosen platform comprises a chassis, four wheels, four PMDC motors and four encoders. The components that are needed to complete the robot system have been chosen. Some components through a selection process and some because they have been available from start of the project. The components chosen are a PC/104 board, Arduino board, 4 H-bridges, a laser range scanner and a linear voltage controller. A dynamic vehicle model has been made from reviewing a range of papers concerning skid-steering vehicles. Two papers were chosen and the dynamic models described in the papers were built. The first dynamic vehicle model showed similar outputs to the outputs given in the paper. The second dynamic vehicle model yielded many problems and the final version still had a problem. The first dynamic vehicle model, that showed the best results, was chosen as the simulation for further use in the project. The laser range scanner was used in an experiment to record the path traveled by the assembled robot system. The recorded data was compared against the chosen dynamic vehicle model. This was done to verify that the model could simulate the behavior of the robot system. A motor model describing the dynamics of the motors was built on background of a series of experiments. The transient behavior of the motor model was verified against recorded data for the motors. Step inputs were used for both model and motors. As the robot was to be an autonomous robot, it was necessary to make a navigation algorithm using the laser range scanner. A selection process yielded the Vector Field Histogram method, that could find the best route for robot by using the laser range scanner data. Controllers were developed for the robot system on background of the information obtained through working with the robot system. The controller was made as a hybrid system controller that allows the systems to change between controllers. A fuzzy logic controller was made to determine motor velocity references when it was driving down a corn field passage. The references differed in magnitude, as the difference determined the rate of angular change for the robot system. If the robot was oriented correctly on the the path the motor velocity references were set to be equal, which made the robot drive in a straight line. The fuzzy logic controller generated references by evaluating feedback data from the laser range scanner. Other controllers were made for the hybrid system controller, though these controllers were open loop and based on simple logic. The simple logic controllers were made to handle situations like obstacles and steering the robot into the next corn field passage. Cascaded inside the hybrid system controller are four PID controllers, one for each motor. The PID controllers were given references from the hybrid system controller and utilized encoder feedbacks to control the motors. The PID controllers were designed in the final model, that comprised both the dynamic vehicle model and the motor model. The hybrid system controller is still to be installed onto the PC/104 board and the PID controllers are still to be installed onto the Arduino board. When the controllers are installed the next step is to adjust the controllers when driving through a corn field passage. When the controllers are implemented and adjusted the robot will be ready for The Field Robot Event 2011.

A master thesis from Aalborg University

Mark II: Design and Control of an Autonomous Mobile Agricultural Robot

[Mark II: Design og Regulering af en Autonom Mobil Landbrugsrobot]