Neural Network-Aided Extended Kalman Filter for SLAM Problem

Choi, Minyong; Sakthivel, R.; Chung, Wan Kyun

doi:10.1109/robot.2007.363565

Cited by 35 publications

(16 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…EKF SLAM may also be sensitive to uncertainty of robot motion, especially when the uncertainty of the robot heading is large [13]. Moreover EKF SLAM usually assumes Gaussian noise with zero mean and thus cannot generate an accurate feature map when noise is colored or when systematic bias error occurs [14].…”

Section: Analytic Approach To Slammentioning

confidence: 99%

See 1 more Smart Citation

A new approach to simultaneous localization and map building with implicit model learning using neuro evolutionary optimization

Kang

Kim

et al. 2010

Appl Intell

View full text Add to dashboard Cite

This paper presents Neuro-Evolutionary Optimization SLAM (NeoSLAM) a novel approach to SLAM that uses a neural network (NN) to autonomously learn both a nonlinear motion model and the noise statistics of measurement data. The NN is trained using evolutionary optimization to learn the residual error of the motion model, which is then added to the odometry data to obtain the full motion model estimate. Stochastic optimization is used, to accommodate any kind of cost function. Prediction and correction are performed simultaneously within our neural framework, which implicitly integrates the motion and sensor models. An evolutionary programming (EP) algorithm is used to progressively refine the neural model until it generates a trajectory that is most consistent with the actual sensor measurements. During this learning process, NeoSLAM does not require any prior knowledge of motion or sensor models and shows consistently good performance regardless of the robot and the sensor noise type. Furthermore, NeoSLAM does not require the data association step at loop closing which is crucial in most other SLAM algorithms, but can still generate an accurate map. Experiments in various complex environments with widely-varying types of noise show that the learning capability of NeoSLAM ensures performance that is consistently less sensitive to noise and more accurate than that of other SLAM methods.

show abstract

Section: Analytic Approach To Slammentioning

confidence: 99%

“…Recent attempts to overcome the limitations of analytic SLAM include using computational intelligence techniques, including Neural Nets (NN) [14], neuro-fuzzy [23], fuzzy logic [24] and genetic algorithms (GA) [25]. Choi et al [14] proposed a neural network-aided EKF algorithm for SLAM.…”

Section: Computational Intelligence Approach To Slammentioning

confidence: 99%

A new approach to simultaneous localization and map building with implicit model learning using neuro evolutionary optimization

Kang

Kim

et al. 2010

Appl Intell

View full text Add to dashboard Cite

show abstract

“…The work presented in this paper is closely related to the Simultaneous Localisation and Mapping (SLAM) [16] problem in which a robot has to build map while at the same time estimating its position relative to the map. Most of the approaches make use of Kalman Filter (KF) and Extended Kalman Filter (EKF) [6,13], [3,4]. Under these methods a matrix representing the robot and landmarks' position is established.…”

Section: Related Workmentioning

confidence: 99%

Localisation and Mapping Using a Laser Range Finder: A Goal-Seeking Approach

Diamantas

Crowder

2009

2009 Fifth International Conference on Autonomic and Autonomous Systems

View full text Add to dashboard Cite

In this paper we examine the problem of localisation and mapping of an unknown environment using data from a laser range finder. In order to support our method we detect landmarks in the environment using the same laser finder. For the localisation and mapping process to take place we assume that the mobile robot will follow a path until a landmark is observed by the laser scanner. Our approach alleviates the requirement to provide odometry or other information. In addition, an efficient path is sought to reach target location. An inherent property of this is obstacle avoidance. The simulated experiments presented in the paper validate the effectiveness of our approach.

show abstract

“…The work in [4] uses Artificial Neural Networks to model the noise in the motion and sensing of a mobile robot, showing improvements over the accuracy of positioning. The main handicap of their approach is that the ANN training must be performed off-line, so the environment has to be well defined, as well as the robot characteristics.…”

Section: Non-linear Enhancement Of Kfmentioning

confidence: 99%

Review of Hybridizations of Kalman Filters with Fuzzy and Neural Computing for Mobile Robot Navigation

Graña

Villaverde

Guede

et al. 2009

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Abstract. Kalman Filters (KF) are at the root of many computational solutions for autonomous systems navigation problems, besides other application domains. The basic linear formulation has been extended in several ways to cope with non-linar dynamic environments. One of the latest trend is to introduce other Computational Intelligence (CI) tools, such as Fuzzy Systems or Artificial Neural Networks inside its computational loop, in order to obtain learning and advanced adaptive properties. This paper offers a short review of current approaches.

show abstract

Neural Network-Aided Extended Kalman Filter for SLAM Problem

Cited by 35 publications

References 14 publications

A new approach to simultaneous localization and map building with implicit model learning using neuro evolutionary optimization

A new approach to simultaneous localization and map building with implicit model learning using neuro evolutionary optimization

Localisation and Mapping Using a Laser Range Finder: A Goal-Seeking Approach

Review of Hybridizations of Kalman Filters with Fuzzy and Neural Computing for Mobile Robot Navigation

Contact Info

Product

Resources

About