Xinzheng Zhang

Sensor fusion of monocular cameras and laser rangefinders for line-based simultaneous localization and mapping (SLAM) tasks in autonomous mobile robots

This paper presents a sensor fusion strategy applied for Simultaneous Localization and Mapping (SLAM) in dynamic environments. The designed approach consists of two features: (i) the first one is a fusion module which synthesizes line segments obtained from laser rangefinder and line features extracted from monocular camera. This policy eliminates any pseudo segments that appear from any momentary pause of dynamic objects in laser data. (ii) The second characteristic is a modified multi-sensor point estimation fusion SLAM (MPEF-SLAM) that incorporates two individual Extended Kalman Filter (EKF) based SLAM algorithms: monocular and laser SLAM. The error of the localization in fused SLAM is reduced compared with those of individual SLAM. Additionally, a new data association technique based on the homography transformation matrix is developed for monocular SLAM. This data association method relaxes the pleonastic computation. The experimental results validate the performance of the proposed sensor fusion and data association method.

A Robust Regression Model for Simultaneous Localization and Mapping in Autonomous Mobile Robot

Segment-based maps as sub-class of feature-based mapping have been widely applied in simultaneous localization and map building (SLAM) in autonomous mobile robots. In this paper, a robust regression model is proposed for segment extraction in static and dynamic environments. We adopt the MM-estimate to consider the noise of sensor data and the outliers that correspond to dynamic objects such as the people in motion. MM-estimates are interesting as they combine high efficiency and high breakdown point in a simple and intuitive way. Under the usual regularity conditions, including symmetric distribution of the errors, these estimates are strongly consistent and asymptotically normal. This robust regression technique is integrated with the extended Kalman filter (EKF) to build a consistent and globally accurate map. The EKF is used to estimate the pose of the robot and state of the segment feature. The underpinning experimental results that have been carried out in static and dynamic environments illustrate the performance of the proposed segment extraction method.

A Comparative Study of Three Mapping Methodologies

Map building is one of the core competencies of truly autonomous robots. Numerous techniques have been developed to represent the static and dynamic environments as well as the perceptional sensing frameworks so far. In this paper, on the basis of our previous work, we compare various sensor systems in building the static and dynamic environment map with the segment-based map and Fuzzy-Tuned Grid-Based Map (FTGBM) strategies. From the comparative results of experiments, we propose a probably efficient and trade-off framework which balances the accuracy of the map against the overall system cost.

A novel mapping strategy based on neocortex model: Pre-liminary results by hierarchical temporal memory

Bio-inspired mapping methods have started a new trend in the robotics navigation area. In this paper, we propose a new map building framework based on the neocortex model: Hierarchical Temporary Memory (HTM). HTM has tree-shaped hierarchical structure and demonstrates structural and algorithmic properties of the human brain neocortex. We first treat the mapping problem as the object recognition problem, and design HTM network hierarchical structure. Secondly, the Speed Up Robust Features (SURF) descriptors were extracted from the grabbed images. These descriptors were further projected into visual words. The presence or absence of visual words consists of input data of HTM in the form of binary sequences. With the binary visual words sequences, HTM network stored or recognized the scene information which were reflected in the visual words, and the output of HTM was the related environment map. After training the HTM network, we evaluated it by two sets of environment data. The results show that the HTM based mapping strategy can build the environment map successfully and handle the loop closing problem with high performance.

Entropy based robust estimator and its application to line-based mapping

This paper presents a robust mapping algorithm for an application in autonomous robots. The method is inspired by the notion of entropy from information theory. A kernel density estimator is adopted to estimate the appearance probability of samples directly from the data. An Entropy Based Robust (EBR) estimator is then designed that selects the most reliable inliers of the line segments. The inliers maintained by the entropy filter are those samples that carry more information. Hence, the parameters extracted from EBR estimator are accurate and robust to the outliers. The performance of the EBR estimator is illustrated by comparing the results with the performance of three other estimators via simulated and real data.

Sensor Fusion for SLAM Based on Information Theory

We present a sensor fusion management technique based on information theory in order to reduce the uncertainty of map features and the robot position in SLAM. The method is general, has no extra postulated conditions, and its implementation is straightforward. We calculate an entropy weight matrix which combines the measurements and covariance of each sensor device to enhance reliability and robustness. We also suggest an information theoretic algorithm via computing the error entropy to confirm the relevant features for associative feature determination. We validate the proposed sensor fusion strategy in EKF-SLAM and compare its performance with an implementation without sensor fusion. The simulated and real experimental studies demonstrate that this sensor fusion management can reduce the uncertainty of map features as well as the robot pose.

An Optimal Graph Theoretic Approach to Data Association in SLAM

In this paper, we study the problem of data association in simultaneous localization and mapping (SLAM). Since almost all existing methods for solving the problem are only able to provide suboptimal solutions, we revisit this problem and propose an optimal graph approach to resolve it. We first formulate the problem as integer programming (IP) problem, and then algorithmically prove that the IP is equivalent to a minimum weight bipartite perfect matching problem. Thus, optimally solving the bipartite matching problem is equivalent to optimally resolve the IP problem (i.e., the data association problem). Simulations validate the effectiveness and accuracy of the proposed approach.

A Virtual Range Finder based on Monocular Vision System in Simultaneous Localization and Mapping

Abstract This paper presents a virtual range finder model with the monocular vision system for simultaneous localization and mapping (SLAM). It relaxes the constraint often cited in the literature that the motion of the optical axis has to be parallel, and reduces the errors for range extraction by a single camera. This model could also provide a supplementary range measurement for landmark initialization in bearing-only SLAM. As the sensor data transformation from pixel to metric value is a nonlinear process, the uncertainty for observation model adopted in Extended Kalman Filter (EKF) SLAM framework can not be in the Gaussian form, which probably makes difficult for data association and SLAM. Concerning this problem, we present a new data association technique based on the homography transformation by a sequence of images and integrate it into the update process of the EKF to assist the innovation computation. The experimental results on real data validated the performance of the virtual range finder model and the new data association approach.

An optimal data association method based on the minimum weighted bipartite perfect matching

Data association is an important problem in simultaneous localization and mapping, however, many single frame based methods only provide suboptimal solutions. In this paper an optimal graph theoretic approach is proposed. We formulate the data association as an integer programming (IP) and then prove that it is equivalent to a minimum weight bipartite perfect matching problem. Therefore, optimally solving the bipartite matching problem implies optimally resolving the IP, i.e. the data association problem. We compare the proposed approach with other widely used data association methods. Experimental results validate the effectiveness and accuracy of the proposed approach, and manifest that this graph based data association method can be used for online application.

An Entropy Optimization Strategy for Simultaneous Localization and Mapping

We present a novel algorithm for simultaneous localization and mapping via application of entropy on construction of segment-based maps. Entropy has been incorporated in SLAM to enhance its sensitivity and robustness in presence of non-Gaussian uncertainties and disturbances. The kernel density estimator is employed to approximate the probability appearance of samples directly from sensor data. An entropy based robust estimator is then designed to extract reliable parameters of the line segment from the environment. Rao–Blackwellized particle filter is also adopted to estimate the pose of the robot and update the map simultaneously. Simulations and experiments results validate the effectiveness and accuracy of the proposed approach.