부산대학교 도서관

Simultaneous localization and mapping (SLAM) is computational technique for robotic system with which it moves in fixed or predefined map having unknown environment. The SLAM has an objective toward localization for the robot in unknown environment of given or predefined map. Many learned people have defined their own way of defining and designing SLAM algorithm in which mSLAM, vSLAM, FullSLAM and Extended Kalman filter (EKF)-based Gmapping SLAM are prominent. Out of which, we found that EKF-based SLAM algorithm performs better. It can be scaled for better précising by adds-on with other curve algorithm. A mixture or serial operation of these algorithms may lead to better optimization of the system. After doing experiments, we stuck with optimizing or creating better precision for localizing in the environment by applying few parametric curve algorithm over EKF. To make it more optimized, now we are experimenting EKF with machine learning (ML)-based optimization for SLAM. SLAM algorithm enables computer systems predict and update robot poses when robot is moving into given map and helps in localization. But these were not very precise in nature due to mathematical approximation of prediction, to overcome it a ML approach may be applied for better precision. In this chapter, we will discuss artificial neural network (ANN)-, k-nearest neighbor (kNN)-, CNN-based optimization techniques for optimizing precious in localization applied on EKF-based SLAM algorithm. To improve precision, an argumentation of ML can be examined, in this respect, we are designing and testing a ML system based on EKF. The system incorporates with argumentation and learning modules based on ML and deep learning. To evaluate the effectiveness of the proposed learning to prediction model, we have developed the ANN-based learning module. For the model trained with deep ML modules, the machine readjusts itself during navigation of robot on different surfaces (marble flooring, boulder tile, rough tile) due to which friction gets controlled and get better path prediction results. Various optimization algorithms are available in deep learning among which kNN, ANN is prominent one which is in great use. Out of which ANN is shallow network which is able to track complex SLAM problem and latter one has lower accuracy for tracked mobile robots. The deep learning system based approach shows significant increase in performance.