Using modified Dijkstra's algorithm to improve the movement efficiency of robocar
| Autor: | Sheng-Yi Wu, 吳昇益 |
|---|---|
| Rok vydání: | 2013 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 101 In recent years, the telehealthcare is very popular. Because the tele-healthcare can keep a watchful eye on information of patients or elderly people, and handle in anytime, in anywhere and by any device, If becomes an on going nursing behavior. Based on its concept, we builded a indoor positioning system by RFID Cartesian grids, which can guide the robocar move to the designated location, then to realize the circumstances with the patient. In this field, many factors will determine whether it can be access or not, such as location-awareness, path finding and path conditions. In this study, we first introduce passive RFID tags to act as landmarks for solving location-awareness. These landmarks can not only read by robocar to determine present localization but reduce computing time for path finding searching process. For the part of path –planning, we proposed an improved graph-based algorithm for archiving obstacles avoidance and less veers into consideration, to generate an efficient path for navigation. We tested the efficiency of different path finding algorithms with the designated map, included Dijkstra’s algorithm, the collision–free algorithm (CFA) on basis of Dijkstra and our proposed method. In comparison of Dijkstra’s algorithm and CFA approach, Dijkstra’s algorithm could find the shortest path. but easily occur collision; and although CFA approach increase 3% distances, it could ensure keeping up a collision-free condition. Another aspect, in comparison of our proposed approach and CFA approach, our method increase cruising distance then CFA, due to it isn’t a shortest path. However, the aim we adopted veering angles is to emend weighting manner to condition of mobile robocar cruising. And our result proved the ideal shortest path is not minimum time to access destinations in practical environment. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|