Mapping Large Loops with a Single Hand-Held Camera

By Laura A. Clemente, Andrew J. Davison, Ian David Reid, José Neira, Juan D. Tardós. Robotics: Science and Systems Conference, 2007

305 citations recorded by the source corpus.

This paper presents a method for Simultaneous Localization and Mapping (SLAM) relying on a monocular camera as the only sensor which is able to build outdoor, closedloop maps much larger than previously achieved with such input. Our system, based on the Hierarchical Map approach [1], builds independent local maps in real-time using the EKF-SLAM technique and the inverse depth representation proposed in [2]. The main novelty in the local mapping process is the use of a data association technique that greatly improves its robustness in dynamic and complex environments. A new visual map matching algorithm stitches these maps together and is able to detect large loops automatically, taking into account the unobservability of scale intrinsic to pure monocular SLAM. The loop closing constraint is applied at the upper level of the Hierarchical Map in near real-time. We present experimental results demonstrating monocular SLAM as a human carries a camera over long walked trajectories in outdoor areas with people and other clutter, even in the more difficult case of forward-looking camera, and show the closing of loops of several hundred meters.

Subfields: Aerospace Engineering, Computer Vision and Pattern Recognition, Electrical and Electronic Engineering

View Paper: https://www.semanticscholar.org/paper/8f75f2af422ab8b73d5a6175a57a0e4382c69781

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 Electrical and Electronic Engineering

Citations in the corpus

Simultaneous Localization and Mapping: A Survey of Current Trends in Autonomous Driving (2017, unknown): “[133] Mapping large loops Coherent map √ Partly (online) ✗ ✗ Very partly [145] Occupancy grid maps merging Coherent map ✗ Partly (online) ✗ ✗ ✗”
Real-time monocular visual odometry for on-road vehicles with 1-point RANSAC (2009, background): “SLAM has been most often performed with other sensors than regular cameras, however in the last years successful results have been obtained using single cameras alone (see [13], [14], [15], and [16]).”
Monocular vision SLAM for indoor aerial vehicles (2009, background): “A more recent work [10] presented a different approach to mitigate the issues involving long distances by means of map matching.”
ORB-SLAM3: An Accurate Open-Source Library for Visual, Visual–Inertial, and Multimap SLAM (2020, methodology): “Mid-term data association was signiﬁcantly improved using techniques that guarantee that the feature matches used are consistent, achieving hand-held visual SLAM [54], [55].”
Incremental smoothing and mapping (2008, background): “[14] presents an interesting approach to visual SLAM that uses the”
Robust reconstruction of indoor scenes (2015, background): “Fragments are analogous to submaps, which are used in a number of robotic mapping systems [25, 5, 54, 10].”
1-Point RANSAC for extended Kalman filtering: Application to real-time structure from motion and visual odometry (2010, background): “This problem can be solved by dividing the large global map to smaller maps with mutual position information (Estrada et al., 2005; Clemente et al., 2007; Bosse et al., 2004; Williams et al., 2009).”
Active Matching (2008, methodology): “At low levels of motion model uncertainty, mismatches via this method are relatively rare, but in advanced applications of the algorithm [1,13] it has been observed that Joint Compatibility testing finds a significant number of matching errors and greatly improves performance.”
Scale Drift-Aware Large Scale Monocular SLAM (2010, methodology): “One of the first attempts to put a whole large scale monocular SLAM together was by Clemente et al. [4], who took a hierarchical mapping approach where locally filtered submaps were built and joined by an upper level joint map.”
Visual SLAM: Why filter? (2012, unknown): “These results led to a series of sub-mapping techniques [7, 17, 43] which are not only motivated by the inconsistencies in filters once uncertainty is large but also by the fact that a filter’s cost increases, typical quadratically, with the map size.”
RGB-D mapping: Using Kinect-style depth cameras for dense 3D modeling of indoor environments (2012, methodology): “2006; Konolige and Agrawal 2008), monocular cameras (Clemente et al. 2007), and even unsorted collections of photos (Snavely et al.”
Fast and Incremental Method for Loop-Closure Detection Using Bags of Visual Words (2008, methodology): “Loop-closure detection has also been performed using an extended Kalman ﬁlter (EKF) application to visual SLAM [16], [17].”
Mapping a Suburb With a Single Camera Using a Biologically Inspired SLAM System (2008, background): “Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org.”
Visual simultaneous localization and mapping: a survey (2012, methodology): “Loop closure detection consists in recognizing a place that has already been visited in a cyclical excursion of arbitrary length (Ho and Newman 2007; Clemente et al. 2007; Mei et al. 2010).”