Embedded-system Design And Research On Path-planning Arithmetic For Mobile Robots

With the development of society, the number of industrial robot has increased. By the newest prediction, the production of home robot occupies the main part of robot-made field by the 21st century. Industrial robots are mainly used in the second industry represented by factories, while home robots usually appear in the service industry. So it has potential markets in normal families. Embedded intelligent system for home mobile robots is deeply researched in this thesis, and it has drew valuable conclusions on arithmetic and hardware design. In this thesis the robot can walk automatically, avoid obstacles in time, search for charger to charge in case of low voltage, and cover all the room. In order to achieve these aims, the following path-planning strategy and controlling arithmetic are adopted in this thesis.The environment of home mobile robot is uncertain (such as different positions of inner objects), so this requires robot should have the ability to adapt to variable surroundings. Genetic Programming has many characters such as powerful ability of solving problems and variable length of genetic coding. And these merits make GP suitably evolve the behaviors of robot. So GP is used to avoid obstacles in the thesis, and the arithmetic of walking across over is used to cover all the room. Another important point is that GP needs less memory, so it is easily implemented on hardware devices.Mobile robot should avoid quickly in case of obstacles and must have powerful ability of processing data. So in order to achieve these aims, hardware devices must have quick processing speed and can also be extended. Based on these requirements, FPGA is used to "carry" GP arithmetic in this thesis and 16-bit MCU SPCE061A is applied to control external circuits to move two motors and change the walking modes. What is more, this MCU has powerful ability of processing voice. And it is very convenient to control actions of robot by voice.This thesis thoroughly depicts the evolving procedures of GP how to evolve robot, including the structures of individuals, genetic operation and fitness. Several problems on arithmetic implementation on FPGA are discussed, which are registers, acquisition of random numbers, crossover operation and efficient memory management.The results of simulation and actual experiments prove that the robot can evolve quickly. First, it walks around the room, and then walk across over. When it comes across obstacles, it can avoid in time. And when it detects low voltage, it begins to search for charger to charge.