Crawler Underground Exploration Robot Low-level Control System Research And Design,

China as a great nation producing the coal in the world, more mine accident happened in the country, needs extremely to take care of mine security. Before the accident happened under the ground, how to detect the danger exactly and duly and to prevent the hidden accident, after the accident how to succor as much as possible, will be the great significance topic. In this article, we will describe the designation of a control system for a motion robot platform which will be suitable for exploring and detecting under mine.At the start of the thesis, there is an introduction of the topic, including the origin, the background, the significance and the existing research status domestic and foreign. Then the article gives an introduction of the mechanism of the robot in order to make a direct-viewing for the robot platform for the readers. There is an introduction for the topic design plan and the target also.The control system of mobile robot platform is a distributional control system based on CAN Bus. In the article, the designation is divided into three parts - the hardware, the algorithm and the programming - to be described. At the same time, formulate a set of command system based on the robot behavior. Through teaching and the rules in the Expert System of the robot own, the robot can optimize the behavior. After the behavior is confirmed, robot can complete the task by using the behaviors.The CAN Bus node is using MC9S12DG128B - FreeScale 16Bit MCU. Besides communications, the MCU is as the basic controller of the robot platform. The robot is drove by power of motors. In the power system, we use SA56 and SA60 - H Bridge type PWM amplifier - as DC brush type motor driver circuit. In the thesis, the structure of the CAN network in the control system is introduced and the functions of every part is explained.In the aspect of algorithm, the robot acts come into being through the artificial demonstration and are packed as behavior objects. A single behavior object contains the algorithm for optimizing the demonstrated group of acts. The algorithm is using the Q-learning based on artificial nerve network. After the robot forms a behavior, it can make its own behavior movement adapting to environment better through the study algorithm. When robot meets the similar environment, the behavior which the robot uses still might be evolved according to the changed environment.In the aspect of programming, there are 4 typical main procedure and subroutines to be introduced in the article. The method of design and the procedure flow are described amply.In the end of the article, summarize the design of the control system according to the actual experimental result. Appraises advantage and disadvantage of this design in the robot prototype, and then describe the possible improving method.