This research project is a combined work of Amulya Khare and Rajul Gupta under the supervision of Asst/P Low Yoke Hean, Malcolm (School of Computer Engineering, Nanyang Technological University). This project was a submission for the Nanyang Research Programme and won a merit award. in the year 2007
Introduction
Civil-military operations such as Search and Rescue (SAR) operations require the rescue team to navigate through unknown terrains and environment. Such operations are aimed at finding people who are lost or injured in urban or remote areas which might be difficult to access. In many cases it may not be feasible to use human rescuers due to the hazardous nature of the environment (e.g. presence of bio-chemical materials or bombs). So such operations require the use of human as well as software controlled robotic agents which may efficiently carry out the required task without much risk to human lives. In operations where large numbers of robots are deployed, these robots have to be controlled by software agents and use different strategies for path-planning to complete the SAR operation quickly and effectively. Such tactics include implementation of different kinds of searches by different robots and co-ordination between them to achieve a common goal. The testing of the efficiency of such robots in different terrains, if done in real world environment, may result in the damage of the robots and may also require resources for the creation of such the test environment. Thus a 3D simulation environment of the terrain can be created to provide a realistic testbed for these software controlled robots. Such an environment is inexpensive and easy to create and provide a fast way to obtain test results. Our project aims at creating a 3D maze as a testbed to demonstrate the efficiency of the different tactics used by these robots in a search and rescue operation of an unknown terrain.
Project Overview
We have used Unreal Tournament (UT), which is a commercial-off-the-shelf (COTS) game system, as a platform to create our 3D maze. UT game engine provides comprehensive tools for the creation of the maze along with advanced visualizations for realistic game play. The UT bots in the maze introduced by JBOT client will be controlled by a Java program that would define its path for the SAR operation in the maze. Three UT bots will be performing three different kinds of searches (Depth-first-search, Breadth-first-search and A*(Star) search) in the same maze simultaneously. The efficiency of these algorithms depends on the map of the terrain and the destination location. By using different maps, the efficiency of the algorithms in different SAR operations can be evaluated.
Project Poster
Project Overview
We have used Unreal Tournament (UT), which is a commercial-off-the-shelf (COTS) game system, as a platform to create our 3D maze. UT game engine provides comprehensive tools for the creation of the maze along with advanced visualizations for realistic game play. The UT bots in the maze introduced by JBOT client will be controlled by a Java program that would define its path for the SAR operation in the maze. Three UT bots will be performing three different kinds of searches (Depth-first-search, Breadth-first-search and A*(Star) search) in the same maze simultaneously. The efficiency of these algorithms depends on the map of the terrain and the destination location. By using different maps, the efficiency of the algorithms in different SAR operations can be evaluated.
Project Poster
0 comments:
Post a Comment