Autonomous Navigation Strategy in Mobile Robot

Abstract

To solve the navigation problem of mobile robot in unknown environment, a navigation scheme based on bionic strategy was proposed, which simulates operant conditioning mechanism. In this scheme, the tendency cell was designed by use of information entropy, which represents the tendency degree for state. The improved Q learning algorithm used as learning core to direct the learning direction. The Boltzmann machine was used to process annealing calculation, which can randomly selected navigation action. The selected strategy of action will tend to optimal with the learning process. Simulation analyses were carried out in mobile robot; results showed that the proposed method had quick learning velocity and accurate navigation ability, and robot could successfully evade obstacles and arrived at goal point with optimal path.

Visual Loop Closure Detection with Scene Mutual Information for Mobile Robot

“…This technology can be applied to the robot navigation in large-scale unstructured environment (shown in Fig. 1 a), which is currently a hot research topic [2][3][4][5]. During the process of visual SLAM, the mobile robot keeps judging whether it is in a position that has been visited previously, so as to ensure the correctness of the environment topological map, i.e., the visual loop-closure detection [6].…”

A Fuzzy Path Preview Algorithm for the Rice Transplanting Robot Navigation System

“…Research on path preview control has been conducted since the 1960s [9]. However, there are lots of differences of the Path Preview algorithms between offroad vehicle and indoor mobile robot [10,11] and industrial robot [12]. Peng, H. et al have proposed the preview control method combined with front feed and feedback and introduced path curvature into the front feed circuit to improve the path trace effect of the entire system [9].…”