Visual Odometry : Part II: Matching, Robustness, Optimization, and Applications

By Friedrich Fraundorfer, Davide Scaramuzza. IEEE robotics & automation magazine, 2012

642 citations recorded by the source corpus.

Subfields: Aerospace Engineering, Computer Vision and Pattern Recognition, Ocean Engineering

View Paper: https://www.semanticscholar.org/paper/aca864792bb96b0258f638fc216741b5a6a8ca9e

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Aerospace Engineering

In Computer Vision and Pattern Recognition

In Ocean Engineering

Citations in the corpus

Multiple‐Robot Simultaneous Localization and Mapping: A Review (2016, methodology): “Various features, such as scale-invariant feature transform (SIFT) and speeded up robust features (SURF), have been introduced in (Fraundorfer & Scaramuzza, 2012).”
Simultaneous Localization and Mapping: A Survey of Current Trends in Autonomous Driving (2017, neutral): “In [89] and [90], the authors cover the topic of Visual Odometry (VO) where the focus is on the localization (and the trajectory) and not on the map.”
Evolution of Indoor Positioning Technologies: A Survey (2017, background): “VO is not a recent approach and has been around for some 30 years but has been gradually improved both in efficiency and in precision [81].”
Visual SLAM algorithms: a survey from 2010 to 2016 (2017, background): “According to the survey papers in robotics [18, 19], the relationship between vSLAM and VO can be represented as follows.”
An Overview to Visual Odometry and Visual SLAM: Applications to Mobile Robotics (2015, background): “In general, the cost function to be minimized can be formulated as [45]: arg min where X i is the i th 3D point, p is its corresponding image point observed in the k th frame and g ( X i , C k ) is the re-projection function according to the camera pose C k .”
Optic-Flow Based Control of a 46g Quadrotor (2013, methodology): “However, methods based on feature tracking remain relatively complex because they require cameras with good enough resolution and sufficient computing power for the feature extraction and matching algorithms [12].”
Visual SLAM and Structure from Motion in Dynamic Environments (2018, background): “Under RANSAC, the number of iterations required to guarantee a correct solution is where s is the number of data points, ε is the percentage of outliers, and p is the probability of success (confidence level) [39].”
DeepVO: Towards end-to-end visual odometry with deep Recurrent Convolutional Neural Networks (2017, background): “1, a classic pipeline [1], [2], which typically consists of camera calibration, feature detection, feature matching (or tracking), outlier rejection (e.”