Development of behavior-based reactive navigation system for mobile robot

• Main Author: Mohd Hamami, Mohd Ghazali
• Format: Thesis
• Language: English
• Published: 2013
• Online Access: http://psasir.upm.edu.my/id/eprint/47585/1/FK%202013%2062R.pdf; http://psasir.upm.edu.my/id/eprint/47585/

Robotics technology has been evolved rapidly these last two decades especially in autonomous mobile robots development. One of the most important issues that related to autonomous mobile robots is it navigation systems. Deliberative navigation and reactive navigation are two types of navigation in mobile robot navigations. Reactive navigation in unknown and changing territories without prior knowledge to the environment is one of the most challenging problems in robotics, thus has been investigated by researchers for many years. Previous researchers gave particular attention to local minimum and multiple minimum problems which trap the robot into an infinite loop during its navigation. This could ruin the objective of good local navigation which a good local navigation should able the robot to navigate the unknown environment safety without having collision with the available obstacles. To solve the problems and achieve the above objective, a new behavior-based reactive navigation system for mobile robots is developed. The motions of Pioneer 3TM mobile robot were simulated to show the developed algorithm performance. The robot equipped with three rendering sensors and eight sonar sensors to senses and perceives its environment. The robot environment consist of wall obstacles and dead end traps such as simple obstacle corner, U shape dead ends, snail shape, loops and maze which all of it, is assumed to be fully unknown. The navigation algorithm consist of four main behaviors which target seeking behavior,obstacles avoidance behavior, tracking (wall following and edge following) behavior and emergency stop behavior. All these four main behaviors have been integrated into one complete behavior-based navigation system using competitive integration methods (winner take all network). This developed behavior-based approach, not only enable the robot to search for the target but it also able the robot to escape from any sort of multiple dead end traps. The algorithm also control the robot velocity during its navigation based on the Euclidean distance of the robot and the target and also distance input from the sensors. In this work, multiple traps may have many types of shape ranges from a simple obstacle corner, U shape dead ends, snail shape until to a complicated loops and mazes. The developed algorithm control the mobile robot and resulted the mobile robot makes optimum logical trajectories towards the target, escape from the dead end traps by avoiding available obstacles and control the robot speed throughout the navigation process. The obtained simulation results were compared with the previous methods to show the effectiveness and efficiency of the proposed method in term of the navigation path pattern result and its algorithm consideration. The repeatability tests shown that the developed navigation algorithm able the robot to reach the target with the similar navigation path pattern outcome in each simulation tests. This proved the developed navigation algorithm not only satisfied its objective but also has reliability in its performance.