Visual Odometry System Using Multiple Stereo Cameras and Inertial Measurement Unit

Oskiper, Taragay; Zhu, Zhiwei; Samarasekera, Supun; Kumar, Rakesh

doi:10.1109/cvpr.2007.383087

Cited by 84 publications

(68 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This is better than the loop closure error obtained in [3] for a combined indoor-outdoor path with a more elaborate system made of four cameras and an IMU. Future work involves bundle adjustment and automatic loop closure detection.…”

Section: Discussionmentioning

confidence: 73%

See 1 more Smart Citation

Image Augmented Laser Scan Matching for Indoor Localization

Naikal¹,

Kua

Zakhor

2009

View full text Add to dashboard Cite

Abstract-Indoor localization is a challenging problem addressed extensively by both robotics and computer vision communities. Most existing approaches focus on using either cameras or laser scanners as the primary sensor for pose estimation. In laser scan matching based localization, finding scan point correspondences across scans is a challenging problem as individual scan points lack unique attributes. In camera based localization, one has to deal with images with little or no visual features as well as scale factor ambiguities to recover absolute distances. In this paper, we develop a multimodal approach for indoor localization by fusing a camera and a laser scanner in order to alleviate the drawbacks of each individual modality. Specifically, we use structure from motion to estimate the pose of a moving camera-laser rig which is subsequently used to compute scan correspondence estimates which are refined using a window based search method for scan point projections on the images. We have demonstrated our proposed system, consisting of a laser scanner and a camera, to result in a 0.4% loop closure error for a 60m loop around the interior corridor of a building. I. INTRODUCTIONocalization in environments with limited global positioning information remains a challenging problem. Indoor localization is a particularly important problem with a number of applications such as indoor modeling, and human operator localization in unknown environments. Localization has been primarily studied by the robotics and computer vision research communities. In robotics, the focus has been on estimating the joint posterior over the robot's location and the map of the environment using sensors such as wheel encoders, laser scanners and Inertial Measurement Units (IMUs). This problem is typically referred to as Simultaneous Localization and Mapping (SLAM) [5]. To localize a wheeled robot, simple 2D maps are typically generated using 2D horizontal laser scanners which serve to both localize the robot and measure depth to obstacles directly. Laser scan matching based localization approaches involve computing the most likely alignment between two sets of slightly displaced laser scans. The Iterative Closest Point (ICP) algorithm [1] is the most extensively used scan correlation algorithm in robotics. The open loop nature of the pose integration from ICP and wheel odometry tends to introduce large drifts in the navigation estimates.This research project is supported by the ARO grant W911NF-07-1-0471 and the HSN MURI grant W911NF-06-1-0076.These estimates can be improved by using a dynamic motion model for the robot, and by applying probabilistic methods to estimate the robot's location and the map. Thrun et. al.[7] use an expectation maximization approach to solve the SLAM problem on robots with wheel encoders and laser scanners. Other approaches to SLAM tend to rely heavily on Extended Kalman Filters (EKFs) [6], since the process of tracking the robot's pose and position of the geometric features in the scene can be elegantly repres...

show abstract

Section: Discussionmentioning

confidence: 73%

“…al. [3] have reported on a more elaborate system consisting of two stereo camera pairs and an IMU to obtain between 0.5% to 1% loop closure error. Mourikis and Roumeliotis [4] achieve a 0.3% loop closure error with a camera-IMU system in a car moving outdoors at much higher speeds than our indoor system.…”

Section: Introductionmentioning

confidence: 99%

Image Augmented Laser Scan Matching for Indoor Localization

Naikal¹,

Kua

Zakhor

2009

View full text Add to dashboard Cite

show abstract

“…This scale is generally determined using GPS waypoints, which makes it inapplicable to indoor environments unless objects of known size are placed in the scene. To resolve this unknown scale factor, stereo camera based approaches have gained popularity, as the extrinsic calibration between the cameras can be used to recover absolute translation parameters [15,16,3]. Se et.…”

Section: Introductionmentioning

confidence: 99%

Image augmented laser scan matching for indoor dead reckoning

Naikal

Kua

Zakhor

2009

2009 IEEE/RSJ International Conference on Intelligent Robots and Systems

View full text Add to dashboard Cite

Abstract-Most existing approaches to indoor localization focus on using either cameras or laser scanners as the primary sensor for pose estimation. In scan matching based localization, finding scan point correspondences across scans is challenging as individual scan points lack unique attributes. In camera based localization, one has to deal with images with few or no visual features as well as scale factor ambiguities to recover absolute distances. In this paper, we develop multimodal approaches for two indoor localization problems by fusing a camera and laser scanners in order to alleviate the drawbacks of each individual modality. For our first problem we recover 3 Degrees of Freedom (DoF) of a camera-laser rig on a rolling cart in a 2D plane, by using visual odometry to facilitate scan correspondence estimation. We demonstrate this approach to result in a 0.3% loop closure error for a 60m loop around the interior corridor of a building. In our second problem, we recover 6 DoF of a human operator carrying a backpack system mounted with sensors in 3D, by merging rotation estimates from scan matching and translation estimates from visual odometry, resulting in a 1% loop closure error.

show abstract

“…Since GPS is susceptible to interference or even not fully available (tunnels, indoor environments, mountainous forested environments), odometry based localization is taken as a supplement solution to calculate the position of the vehicle by using e.g.vision based sensors (cameras, LIDARs, etc) [1], wheel sensors [2] and inertial sensors (gyroscopes and accelerometers) [3] [4]. However, as an odometry system without references to re-calibrate itself, the position will drift with the passage of time.…”

Section: Introductionmentioning

confidence: 99%

A sensor fusion approach for localization with cumulative error elimination

Zhang

Stähle

Chen³

et al. 2012

2012 IEEE International Conference on Multisensor Fusion and Integration for Intelligent Systems (MFI)

View full text Add to dashboard Cite

Abstract-This paper describes a robust approach which improves the precision of vehicle localization in complex urban environments by fusing data from GPS, gyroscope and velocity sensors. In this method, we apply Kalman filter to estimate the position of the vehicle. Compared with other fusion based localization approaches, we process the data in a public coordinate system, called Earth Centred Earth Fixed (ECEF) coordinates and eliminate the cumulative error by its statistics characteristics. The contribution is that it not only provides a sensor fusion framework to estimate the position of the vehicle, but also gives a mathematical solution to eliminate the cumulative error stems from the relative pose measurements (provided by the gyroscope and velocity sensors). The experiments exhibit the reliability and the feasibility of our approach in large scale environment.

show abstract

Visual Odometry System Using Multiple Stereo Cameras and Inertial Measurement Unit

Cited by 84 publications

References 14 publications

Image Augmented Laser Scan Matching for Indoor Localization

Image Augmented Laser Scan Matching for Indoor Localization

Image augmented laser scan matching for indoor dead reckoning

A sensor fusion approach for localization with cumulative error elimination

Contact Info

Product

Resources

About