Navigation New Technology Major trend Allows Exploration of dangerous environments Presents problems of uncertainty and imprecision Presents difficulty structuring the navigation task Behavior Based Navigation Is a form to structure the navigation task Employs the divide and conquer approach Makes the system modular Simplifies navigation solution Can add new behaviors Behavior-Based Control Schemes Types: Sub-sumption architecture arranges the behavior by levels of priority . The motor schemas Function with forces (repulsion, attraction, overall) Fuzzy logic system to represent and coordinate the behaviors. operate under imprecise and uncertain environments Fuzzy Logic Behavior Based Navigation Steps The problem is divided into smaller simpler tasks. Each task will represent a behavior. Each behavior is represented by a set of fuzzy logic rule statements. Example: (1) IF dis VERYNEAR, THEN v is STOP. (2) IF dis NEAR, THEN v is SLOW. Behavior coordination Priorities (weights) Behavior Based Strategies traverse behavior method 2002 minimum risk method 2008 selection behavioral approach 2008 fixed motor schema approach 2008 Traverse Behavior Method This behavior extracts three characteristics from video images: Roughness Slope Discontinuity.
This data is used to create an index to guide the robot
to the safest route. Minimum Risk Method the robot navigation consist of three behaviors Global------ Goal Seeking Regional--- Path Searching Local-------Obstacle Avoidance. Depending on the robots position the weights of GS, OA, and PS changes. Their respected minimum risk method is provided by the path searching behavior. Minimum Risk Method The objective is to avoid encountering obstacles and avoid repeating previous trajectories This is done by assigning a matrix with the respected values OMD--- the risk that the robot will collide with an obstacle TMD---- the risk that the robot might iterate a previous trajectory
These values are then stored on a memory grid map in
which the robot work area is represented in a 2-D array of cells. Minimum Risk Method Examples Selection Behavioral Approach Therefore a behavior selection based navigation and obstacle avoidance algorithm was implemented. The algorithm allows the robot to select the appropriate behavior, based on information provided by the sensors Furthermore the behavior selection approach gives the robot the ability to pick between the different behaviors available, select a different set of behavioral approaches Make some small changes to a behavioral approach Fixed Motor Schema Approach the navigation task is divided into specific motor skills or schemas. A goal is an attractive force An obstacle represents a repulsive force The sum of these forces represents the coordinated action. Set of fuzzy meta rules for each motor schema to scale each force in order to avoid local minima errors. Fixed Motor Schema Approach Conclusion Robot navigation is still a new growing science There are different kinds of methods In general there isn’t a method which solves all the problems in an efficient manner. The efficiency depends on the objectives that each designer decides to attack The difficulty lies when the number of problems attacked increases since the overall efficiency falls.