Enhanced SLAM for a mobile robot using extended Kalman Filter and neural networks

Choi, Kyung‐Sik; Lee, Suk-Gyu

doi:10.1007/s12541-010-0029-9

Cited by 29 publications

(12 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, some of the initiatives also introduce learning approach. For example, Choi and Lee [7] introduces a hybrid approach on SLAM by combining extended Kalman filter and neural networks as to derive accurate solutions to the navigation problem. The results indicate improvement towards the existing SLAM approach.…”

Section: Related Workmentioning

confidence: 99%

Learning above-and-below relationship for vision-based robot navigation system using Distributed Hierarchical Graph Neuron (DHGN) algorithm

Amin

Ahmad

Sayeed

et al. 2014

2014 13th International Conference on Control Automation Robotics &Amp; Vision (ICARCV)

View full text Add to dashboard Cite

Obstacle avoidance is one of the important considerations in developing a vision-based robot navigation system. For flying robots, the ability to learn the above-and-below relationship for obstacle avoidance is necessary. This paper presents a conceptual work in developing a learning mechanism to identify the above-and-below relationship for obstacle avoidance in vision-based robot navigation system using a pattern recognition algorithm known as Distributed Hierarchical Graph Neuron (DHGN). DHGN is a bio-inspired pattern recognition algorithm that implements learning and memorization through a distributed and hierarchical processing. Preliminary results of simple above-and-below navigation with binary images using DHGN indicate that the scheme is able to produce high recall accuracy for obstacle detection. In addition, the proposed scheme implements a one-shot learning approach that is suitable for realtime deployment in robot navigation system.

show abstract

Section: Related Workmentioning

confidence: 99%

Learning above-and-below relationship for vision-based robot navigation system using Distributed Hierarchical Graph Neuron (DHGN) algorithm

Amin

Ahmad

Sayeed

et al. 2014

2014 13th International Conference on Control Automation Robotics &Amp; Vision (ICARCV)

View full text Add to dashboard Cite

show abstract

“…(32). Under the assumption that this process does not have any bias, the n, n1, n2 and n3 describe the number of input nodes, the first hidden layer's nodes, the second hidden layer's nodes and output layer's nodes with A, B and C, the number of nodes, respectively [8].…”

Section: Time Update (Predict)mentioning

confidence: 99%

“…In order to solve SLAM problems, statistical approaches, such as Bayesian Filters, have received widespread acceptance [7]. Some of the most popular approaches for SLAM include using a Kalman filter (KF), an extended Kalman filter (EKF) and an unscented Kalman filter (UKF) and a particle filter [8]. The UKF SLAM makes a Gaussian noise assumption for the robot motion and its observation.…”

Section: Introductionmentioning

confidence: 99%

Hybrid Filter Based Simultaneous Localization and Mapping for a Mobile Robot

Panah

Faez

2011

Image Analysis and Processing – ICIAP 2011

View full text Add to dashboard Cite

Abstract.A mobile robot autonomously explores the environment by interpreting the scene, building an appropriate map, and localizing itself relative to this map. This paper presents a Hybrid filter based Simultaneous Localization and Mapping (SLAM) approach for a mobile robot to compensate for the Unscented Kalman Filter (UKF) based SLAM errors inherently caused by its linearization process. The proposed Hybrid filter consists of a Multi Layer Perceptron (MLP) for neural network and UKF which is a milestone for SLAM applications. The proposed approach, based on a Hybrid filter, has some advantages in handling a robotic system with nonlinear motions because of the learning property of the MLP neural network. The simulation results show the effectiveness of the proposed algorithm comparing with an UKF based SLAM.

show abstract

“…11 Generally, EKFs dealing with nonlinear systems tend to diverge when the initial values and covariances are larger than the actual values. 12,13 In this paper, USAT positioning algorithms are presented. Calculation of USAT positions using the inverse matrix method 7 suffers from singular problems caused by the USAT transmitter layout.…”

Section: Introductionmentioning

confidence: 99%

Ultrasonic satellite system moving object positioning by extended Kalman filter

Yoon

Kim

Choi

2011

Int. J. Precis. Eng. Manuf.

View full text Add to dashboard Cite

This paper presents positioning algorithms for an ultrasonic satellite system (USAT) consisting of multiple ultrasonic transmitters and receivers in buildings. The previously used inverse matrix method of calculating USAT positions suffers from problems related to transmitter layout, and the method is sensitive to sensor noise. To solve these problems, a geometric approach with verification by a comparison of simulations with the inverse matrix method, is suggested. However, when an object is moved quickly, the positioning error is increased. The moving object positioning algorithm with an extended Kalman filter (EKF),which takes account of the dynamics of the moving object, is therefore proposed for estimating USAT positioning during movement. The accuracy of the proposed algorithm is evaluated by simulations and experiments. The experimental results show that the proposed algorithm gives a better performance for dynamic states.

show abstract

Enhanced SLAM for a mobile robot using extended Kalman Filter and neural networks

Cited by 29 publications

References 12 publications

Learning above-and-below relationship for vision-based robot navigation system using Distributed Hierarchical Graph Neuron (DHGN) algorithm

Learning above-and-below relationship for vision-based robot navigation system using Distributed Hierarchical Graph Neuron (DHGN) algorithm

Hybrid Filter Based Simultaneous Localization and Mapping for a Mobile Robot

Ultrasonic satellite system moving object positioning by extended Kalman filter

Contact Info

Product

Resources

About