Generalized-ICP

By Aleksandr V. Segal, Dirk Hähnel, Sebastian Thrun. Robotics: Science and Systems Conference, 2009

2,130 citations recorded by the source corpus.

—In this paper we combine the Iterative Closest Point (’ICP’) and ‘point-to-plane ICP‘ algorithms into a single probabilistic framework. We then use this framework to model locally planar surface structure from both scans instead of just the ”model” scan as is typically done with the point-to-plane method. This can be thought of as ‘plane-to-plane’. The new approach is tested with both simulated and real-world data and is shown to outperform both standard ICP and point-to-plane. Furthermore, the new approach is shown to be more robust to incorrect correspondences, and thus makes it easier to tune the maximum match distance parameter present in most variants of ICP. In addition to the demonstrated performance improvement, the proposed model allows for more expressive probabilistic models to be incorporated into the ICP framework. While maintaining the speed and simplicity of ICP, the Generalized-ICP could also allow for the addition of outlier terms, measurement noise, and other probabilistic techniques to increase robustness.

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

View Paper: https://www.semanticscholar.org/paper/b352b3a7f1068b2d562ba12a446628397dfe8a77

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 Mechanical Engineering

Citations in the corpus

3-D Mapping With an RGB-D Camera (2014, methodology): “[17] who use visual features in combination with GICP [18] to create and optimize a pose graph.”
Dense visual SLAM for RGB-D cameras (2013, background): “This class of algorithms is known as iterative closest point (ICP) [18], and many variations of the algorithm have been explored [19], [20].”
DeepVCP: An End-to-End Deep Neural Network for Point Cloud Registration (2019, background): “The implementations of ICP, G-ICP, AA-ICP, and NDT-P2D are from the Point Cloud Library (PCL) [34].”
Evaluating Egomotion and Structure-from-Motion Approaches Using the TUM RGB-D Benchmark (2012, background): “Different sensor modalities have been explored in the past, including 2D laser scanners [25], [26], 3D scanners [27]– [29], monocular cameras [17], [22]–[24], [30]–[32], stereo systems [33], [34] and recently RGB-D sensors such as the Microsoft Kinect [1]–[3].”
Nonparametric Calibration for Depth Sensors (2015, methodology): “The SLAM system is graph based [3] and uses Generalized ICP [8] to match the point clouds extracted from depth data as a front-end and g2o [5] as optimization backend.”
Fast LIDAR localization using multiresolution Gaussian mixture maps (2015, background): “Many similar robotic applications use iterative closest point (ICP) [9], generalized iterative closest point (GICP) [10], normal distributions transform (NDT) [11], or a similar variant to register an observed point cloud to another point cloud or distribution.”
A Review of Point Cloud Registration Algorithms for Mobile Robotics (2015, background): “With the need for 3D localization, systems based on rotating 2D laser rangefinders recently received an increasing attention for both indoor [Armesto et al., 2010] and outdoor [Segal et al., 2009] environments.”
An evaluation of the RGB-D SLAM system (2012, background): “(ICP) algorithm [2], [26], [27] for aligning pairs of local point clouds to establish constraints between observations.”
LeGO-LOAM: Lightweight and Ground-Optimized Lidar Odometry and Mapping on Variable Terrain (2018, methodology): “Generalized-ICP [4] proposes a method that matches local planar patches from both scans.”
Multiple‐Robot Simultaneous Localization and Mapping: A Review (2016, methodology): “For laser rangers, generalized ICP is used in a 3D scan matching algorithm for 3D laser odometry (Segal et al., 2009).”
Improving RGB-D SLAM in dynamic environments: A motion removal approach (2017, methodology): “For instance, SLAMmethods that use Iterative Closest Point (ICP) [13] algorithm for point-cloud registration normally adopt a two-step framework like themechanism in [14].”
A comprehensive survey on point cloud registration (2021, methodology): “The registration problem has endured thorough investigation from optimization aspects [5], [6], [24], [33], [44], [47], [54], [90], [104].”
HDMapNet: An Online HD Map Construction and Evaluation Framework (2021, methodology): “First, pairwise alignment algorithm like ICP [1], NDT [2] and their variants [3] are employed to match LiDAR data at two nearby timestamps using semantic [4] or geometry information [5].”
NICP: Dense normal based point cloud registration (2015, methodology): “Among all these variants, Generalized ICP (GICP) [3] showed to be one of the most effective and robust.”
A Review of Visual-LiDAR Fusion based Simultaneous Localization and Mapping (2020, methodology): “The work in [57] showed a particle ﬁlter applied to 2D SLAM able to compute a highly accurate proposal distribution based on a likelihood model.”
A robust RGB-D SLAM algorithm (2012, methodology): “For our ICP algorithm we employ the state of the art Generalised ICP [15], which is a type of hybrid point-topoint and point-to-plane implementation that can overcomes limitations present in both methods.”
Complex urban dataset with multi-level sensors from highly diverse urban environments (2019, methodology): “…relative rotation ( R rl ) and translation ( t rl ) were calculated by minimizing the error between the projected points using the GICP algorithm (Segal et al., 2009): Computed transformation maps the i th right LiDAR point cloud data ( p r , i ) to the corresponding left LiDAR point cloud data…”