LSD-SLAM: Large-Scale Direct Monocular SLAM

By Jakob J. Engel, Thomas Schöps, Daniel Cremers. 2014

4,059 citations recorded by the source corpus.

Subfields: Aerospace Engineering, Computer Vision and Pattern Recognition, Geology, Robotics & Control

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 Geology

In Robotics & Control

Citations in the corpus

WoodScape: A Multi-Task, Multi-Camera Fisheye Dataset for Autonomous Driving (2019, methodology): “Some examples of the few datasets that have fisheye are: Visual SLAM ground truth for indoor scenes with omnidirectional cameras in [7], SphereNet [9] containing 1200 labelled images of parked cars using 360° cameras (not strictly fisheye) and, in automotive, the Oxford Robotcar 1 ar X iv :1 90 5.”
Continuous Scene Representations for Embodied AI (2022, background): “Classical mapping and localization work in the robotics literature [16, 19, 35] relies on low-level geometric features and does not encode semantic information.”
3D-R2N2: A Unified Approach for Single and Multi-view 3D Object Reconstruction (2016, neutral): “One typical assumption is that features can be matched across views [3,4,5,6] as hypothesized by the majority of the methods based on SFM or SLAM [7,8].”
(CAD)$^2$RL: Real Single-Image Flight without a Single Real Image (2016, background): “Monocular cameras, on the other hand, require 3D estimation from motion, which remains a challenging open problem despite considerable recent progress [13, 20].”
RLBench: The Robot Learning Benchmark & Learning Environment (2019, neutral): “However, if we would like a manipulation system to make use of a SLAM map, it is not currently clear what the best way to represent this map is: whether it be sparse [54], [55], dense [56], [57], or semidense [58].”
IQA: Visual Question Answering in Interactive Environments (2017, background): “Online map-based methods [10, 13, 50, 61, 67, 69] often construct the map while exploring the environment.”
Learning to fly by crashing (2017, methodology): “Several methods [1], [3], [9]–[13] use range or visual sensors to infer maps of the environment while simultaneously estimating its position in the map.”
Past, Present, and Future of Simultaneous Localization and Mapping: Toward the Robust-Perception Age (2016, neutral): “Semi-dense methods overcome the high-computation requirement of dense method by exploiting only pixels with strong gradients (i.e., edges) [88, 101]; semi-direct methods instead leverage both sparse features (such as corners or edges) and direct methods [100, 101, 140] and are proven to be the most…”
Iterated extended Kalman filter based visual-inertial odometry using direct photometric feedback (2017, methodology): “More recently, by employing highly optimized singleinstruction multiple data implementations, first real-time, CPU-based approaches for semi-dense motion estimation using a monocular camera (Engel et al., 2014; Forster et al., 2014) have been proposed.”
A Review of Visual-LiDAR Fusion based Simultaneous Localization and Mapping (2020, methodology): “In this category, the most known methods are DTAM [25], LSD-SLAM [26], SVO [27], or DSO [28,29].”
Computing Systems for Autonomous Driving: State of the Art and Challenges (2020, background): “For situations that lack computation resources, semiDense [128], [129] SLAM”
High-fidelity simulation for evaluating robotic vision performance (2016, background): “dense [5], [20] or dense [16] reconstruction of environment for visual SLAM and other similar applications.”
VINS-Mono: A Robust and Versatile Monocular Visual-Inertial State Estimator (2017, methodology): “Noticeable approaches include PTAM [1], SVO [2], LSD-SLAM [3], DSO [5], and ORB-SLAM [4].”
Geometry-Aware Learning of Maps for Camera Localization (2017, methodology): “To address this issue, recently, direct [27, 43] and semi-direct methods [18, 20] utilize all the pixels with high gradients rather than features to build maps.”
Aligning Cyber Space With Physical World: A Comprehensive Survey on Embodied AI (2024, neutral): “…data and multi-view geometry to estimate a robot’s pose and construct low-level maps (e.g., sparse, semi-dense, or dense point clouds) through methods like ﬁlter-based approaches (e.g., MonoSLAM [58]), keyframe-based methods (e.g., ORB-SLAM [59]), and direct trackingtechniques(e.g.,LSD-SLAM[60]).”
TartanAir: A Dataset to Push the Limits of Visual SLAM (2020, background): “Impressive progress has been made with geometric-based [2]–[5], learning-based [6]–[8], and hybrid methods [9]–[11].”
Deep Virtual Stereo Odometry: Leveraging Deep Depth Prediction for Monocular Direct Sparse Odometry (2018, methodology): “In recent years, large progress has been achieved in the development of monocular VO and SLAM methods [31,9,8,32].”
BA-Net: Dense Bundle Adjustment Network (2018, methodology): “Following the photometric BA (Engel et al., 2014; 2018), we solve our feature-metric BA using a coarse-to-fine strategy with feature map warping at each iteration.”