Robot Design Optimization by means of a Genetic Algorithm and Physics Simulation
Abstract: This thesis presents a new robot design paradigm that utilizes evolutionary optimization techniques and advanced physics simulations. This technology makes it possible to design and test robots in virtual environments before the physical robots are built, which enables robot manufacturers to improve the performance of their products and decrease the time and cost for development. In this project, a 3D robot model was defined in geometric, kinematic and dynamic terms. Also, a piece of software was developed in C++ to optimize the robot design, and to simulate and visualize the robot model with the aid of a physics engine. A genetic algorithm was developed for the optimization and used to minimize the average positional error and the total torque magnitude under constraints on speed, and the design variables were the PID controller parameters and the torque actuator limits. Only predefined robots can be programmed and simulated with current software packages for offline-programming and robot simulation. It was concluded that such software packages can be improved by robot design optimization using the software developed in this project, by means of a genetic algorithm and simulations using a physics engine.
AT THIS PAGE YOU CAN DOWNLOAD THE WHOLE ESSAY. (follow the link to the next page)