The extended Kalman filter (EKF) has become a popular solution for the simultaneous localization and mapping (SLAM). This paper presents the implementation of the EKF coupled with a feedforward neural network for the monocular SLAM. The neural extended Kalman filter (NEKF) is applied online to approximate an error between the motion model of the mobile robot and the real system performance. Inadequate modeling of the robot motion can jeopardize the quality of estimation. The paper shows integration of EKF with feedforward neural network and simulation analysis of its consistency and implementation of the NEKF with a mobile robot, laboratory experimental environment, and a simple USB camera. The simulation and experimental results show that integration of neural network into EKF prediction–correction cycle results in improved consistency and accuracy.
Leonard J, Durrant-Whyte H and Cox I. Dynamic map building for an autonomous mobile robot. In: IEEE international workshop on intelligent robots and systems IROS '90, Ibaraki, IEEE, pp.89–96, 3–6 July 1990.
2.
FreseU. Interview: Is SLAM solved?Künstliche Intelligenz2010; 24: 255–257.
3.
MiljkovićZVukovićNMitićM. New hybrid vision-based control approach for automated guided vehicles. Int J Adv Manuf Technol2013; 66: 231–249.
4.
BabićBMiljkovićZVukovićN. Towards implementation and autonomous navigation of an intelligent automated guided vehicle in material handling systems. Iran J Sci Technol-Trans B: Eng2012; 36: 25–40.
5.
VukovićNMiljkovićZ. A growing and pruning sequential learning algorithm of hyper basis function neural network for function approximation. Neural Network2013; 46C: 210–226.
6.
VukovićNMiljkovićZ. Robust sequential learning of feedforward neural networks in the presence of heavy-tailed noise. Neural Network2015; 63: 31–47.
7.
Vuković N. Machine learning of intelligent mobile robot based on artificial neural networks. PhD Thesis (In Serbian), University of Belgrade, 2012.
8.
MitićMMiljkovićZ. Neural network learning from demonstration and epipolar geometry for visual control of a nonholonomic mobile robot. Soft Comput2013; 18: 1011–1025.
9.
Miljković Z and Aleksendrić D. Artificial neural networks – Solved examples with theoretical background (in Serbian). Belgrade: University of Belgrade, 2009.
10.
MitićMMiljkovićZ. Bio-inspired approach to learning robot motion trajectories and visual control commands. Expert Syst Appl2015; 42: 2624–2637.
11.
ParhiDRSinghMK. Real-time navigational control of mobile robots using an artificial neural network. Proc IMechE, Part C: J Mechanical Engineering Science2009; 223: 1713–1725.
12.
Al-AssadiHMAAMat IsaAAHasanAT. Development of a real-time-position prediction algorithm for under-actuated robot manipulator by using of artificial neural network. Proc IMechE, Part C: J Mechanical Engineering Science2011; 225: 1991–1998.
13.
AlthoeferKLaraBZweiriYH. Automated failure classification for assembly with self-tapping threaded fastenings using artificial neural networks. Proc IMechE, Part C: J Mechanical Engineering Science2008; 222: 1081–1095.
14.
DiryagAMitićMMiljkovićZ. Neural networks for prediction of robot failures. Proc IMechE, Part C: J Mechanical Engineering Science2014; 228: 1444–1458.
15.
KramerKStubberudS. Analysis and implementation of a neural extended Kalman filter for target tracking. Int J Neural Syst2006; 16: 1–13.
16.
KramerKStubberudS. Tracking of multiple target types with a single neural extended Kalman filter. Int J Intell Syst2010; 25: 440–459.
17.
ZhanRWanJ. Neural network-aided adaptive unscented Kalman filter for nonlinear state estimation. IEEE Signal Process Lett2006; 13: 445–448.
Chong K and Kleeman L. Accurate odometry and error modelling for a mobile robot. In: International conference on robotics and automation, Albuquerque, NM, IEEE, 1997, pp.2783–2788.
20.
Larsen T, Bak M, Andersen N, et al. Location estimation for autonomously guided vehicle using an augmented Kalman filter to autocalibrate the odometry. In: Proceedings of FUSION 98 Spie Conference, 6–9 July 1998, Las Vegas, USA.
21.
MartinelliATomatisNSiegwartR. Simultaneous localization and odometry self calibration for mobile robot. Autonom Rob2007; 22: 75–85.
22.
Kang JG, Choi WS, An SY et al. Augmented EKF based SLAM method for improving the accuracy of the feature map. In: IEEE/RSJ international conference on intelligent robots and systems (IROS), Taipei International Convention Center (TICC), Taipei, 18–22 October 2010, pp.3725–3731. IEEE/RSJ.
23.
Choi M, Sakthivel R and Chung WK. Neural network-aided extended Kalman filter for SLAM problem. In: Proceedings of the IEEE international conference on robotics and automation, Roma, 2007, Italy, pp.1686--1690. IEEE, paper 1-4244-0602-1/07.
24.
Kang JG, An SY and Oh SY. Modified neural network aided EKF based SLAM for improving an accuracy of the feature map. In: International joint conference on neural networks (IJCNN), Barcelona, Spain, 2010, pp.1--7, paper 978-1-4244-6916-1.
25.
CiveraJDavisonAMontielJ. Inverse depth parametrization for monocular SLAM. IEEE Trans Rob2008; 24: 932–945.
26.
CiveraJGrasaODavisonA. 1-point RANSAC for EKF filtering. Application to real-time structure from motion and visual odometry. J Field Rob2010; 27: 609–631.
27.
Bailey T, Nieto J, Guivant J et al. Consistency of the EKF-SLAM algorithm. In: IEEE/RSJ international conference on intelligent robots and systems, Beijing, 9--15 October 2006, pp.3562--3568. IEEE/RSJ.
28.
Bar-ShalomYRong LiXKirubarajanX. Estimation with applications to tracking and navigation: theory algorithms and software, John Wiley & Sons, 2001.
29.
RostenEPorterRDrummondT. FASTER and better: A machine learning approach to corner detection. IEEE Trans Pattern Anal Mach Intell2010; 32: 105–119.
30.
VukovićNMiljkovićZ. New hybrid control architecture for intelligent mobile robot navigation in a manufacturing environment. FME Trans2009; 37: 9–18.
