The mining industry desires to cut costs and to operate in more dangerous mines, which is why companies such as Atlas Copco are developing autonomous vehicles. The problem to navigate autonomously is however complex, so the vehicles have in the recent years become more intelligent; the number of computers, actuators and sensors are increasing. For example, an autonomous LHD (Loading – Hauling – Dump) vehicle has sensors including: inertial measurement unit (IMU), odometer, hinge angle sensor, laser range finders and cameras. The parameters for the sensors needs to be calibrated before the vehicle can be used in a mine. There are also a number of electrical currents which needs to be calibrated for the actuators.The calibration of parameters has traditionally been made manually, but Atlas Copco realizes that manual calibration is not feasible once the sales of intelligent mine vehicles increases. Effort is therefore put into automation of the calibration procedures.Interviews with employees were carried out at Atlas Copco to identify the most time consuming procedures during calibration and installation of autonomous mine vehicles. The calibration of steering currents was not only identified as the most time consuming procedure, but also as one of the most complex procedures.The goal of this thesis is to enable easier and quicker installation of mine vehicles. This is done through investigation of methods for automatic calibration of steering currents. The problem is approached from two angles: a grey box model using system identification and a black box model using neural network with resilient backpropagation. The models are compared to a search algorithm, used for simulation of the manual calibration method. In the end, the models are evaluated with regard to performance and ease of implementation.The hypothesis was that the more complex grey box or black box model would have higher accuracy than a simple search algorithm. However, the search algorithm proves to outperform the other models both with regard to accuracy and calibration time, and is also easier to implement. The search algorithm is thus suggested to be implemented instead of a complex model. Moreover, it is suggested that a straightforward mapping of 20 currents may outperform even the search calibration. It is also concluded that calibration of steering currents can be done when the vehicle is standing still.

The mining industry is constantly under pressure to improve productivity, effciency and safety. Although an increased use of automation technology has the potential of con- tributing to improvements in all three factors mines have been relatively slow to make use of automation technology. Automation in the underground mining environment is a challenging prospect for a number of reasons not least of which being the diffculties and associated costs of installing infrastructure in this hazardous environment. The work described in this dissertation focuses on the modelling of a Load-Haul-Dump or LHD vehicle for the purpose of autonomous navigation and control. Considerable progress has been made in automating underground mining vehicles in recent years, and successful test installations have been made. There are still however a number of shortcomings in the existing autonomous underground mine vehicle navigation systems. This dissertation attempts to address some of these problems through the development of a more accurate vehicle model for an LHD vehicle incorporating some vehicle and tyre dynamics thereby potentially reducing the number of sensors and the amount of installed infrastructure necessary to implement the vehicle navigation system. Simulation results are provided for different vehicle modelling techniques and the results are compared and discussed in terms of their suitability for physical implementation in an underground mine. / Dissertation (MEng (Electronic Engineering))--University of Pretoria, 2007. / Electrical, Electronic and Computer Engineering / MEng / unrestricted