As the application of robotic technology reaches beyond the factory floor into unstructured, dynamic, and populated environments, powerful motion, generation algorithms become essential for safety, versatility, dexterity, and, speed of task execution. This thesis proposes the integration of motion planning and motion execution algorithms as a framework to address this challenge. Two motion generation methodologies are presented. The global dynamic window approach applies to mobile bases with two or three degrees of freedom; the elastic strip approach is a more general framework applicable to manipulators with many degrees of freedom. Both approaches integrate motion planning and motion execution algorithms. This integration results in motion generation approaches that retain desirable global properties of planning method, while maintaining the advantages associated with reactive execution approaches. Experimental results are presented that show both approaches to be highly effective in practice. The application of these approaches results in a significant increase of the robot's capability to move safely and robustly in dynamic environments. The computational complexity of the elastic band framework is analyzed and compared with motion planning algorithms.
展开▼
