Cognitive Robotics 10 PDF

Cognitive Robotics 10 Dr. Umesh K Map Building The phrase map building refers to the construction of a map of the work space autonomously by the robot, which enables the robot to plan the optimal path to the goal. Map building helps the mobile robot to become conversant...

Cognitive Robotics 10 Dr. Umesh K Map Building The phrase map building refers to the construction of a map of the work space autonomously by the robot, which enables the robot to plan the optimal path to the goal. Map building helps the mobile robot to become conversant with the world around it. The information about the neighborhood world of the robot is thus required to be encoded in the form of a knowledge base. For the purpose of navigational planning, a mobile robot must acquire knowledge about its environment.  The acquisition of knowledge is a pertinent factor in the process of building perception. Human beings can acquire knowledge from their environment through a process of automated learning. Machines too can acquire knowledge by sensing and integrating consistent sensory information. A number of techniques are prevalent for automated acquisition of knowledge. The most common among them are unsupervised and reinforcement learning techniques. Constructing a 2D World Map It is assumed that the height of the robot is less than that of the obstacles within the workspace. In fact, most of the navigational problems for robots are confined to two-dimensional environments. There exist two different types of algorithms for automated map building. Landmark-based map building Metric-based map building Data Structure for Map Building Let us consider a circular mobile robot that can orient itself in any of the following eight directions: north (N), north-east (NE), east (E), south-east (SE), south (S), south-west (SW), west (W) and northwest (NW).  The process is thus continued until all the obstacles are visited. Another point needs to be noted here, that after moving to an obstacle, the robot should move around it to identify the boundary of the obstacles. Thus when all the obstacles are visited a map representing the boundary of all obstacles will be created. This map is hereafter referred to as the 2D world map of the robot. Two procedures are given below, i.e. Map Building and Traverse Boundary.  In the procedure Map Building a linked list structure is used with four fields. The first two fields, xi, yi denote the coordinate of the point visited by the robot. The third field points to the structures containing the obstacle to be visited next, and the fourth field denotes the pointer to the next point to be visited on the same obstacle.  Another data structure is used in procedure Map Building for acquiring the boundary points visited around an obstacle. This structure has three fields, the first two correspond to the xi, yi coordinate of one visited point on the obstacle i, while the third field is a pointer which corresponds to the next point on obstacle i. An algorithm for map building is presented below, where the procedure Traversal Boundary has been utilized.