SVO: Fast semi-direct monocular visual odometry

Abstract

We propose a semi-direct monocular visual odometry algorithm that is precise, robust, and faster than current state-of-the-art methods. The semi-direct approach eliminates the need of costly feature extraction and robust matching techniques for motion estimation. Our algorithm operates directly on pixel intensities, which results in subpixel precision at high frame-rates. A probabilistic mapping method that explicitly models outlier measurements is used to estimate 3D points, which results in fewer outliers and more reliable points. Precise and high frame-rate motion estimation brings increased robustness in scenes of little, repetitive, and high-frequency texture. The algorithm is applied to microaerial-vehicle state-estimation in GPS-denied environments and runs at 55 frames per second on the onboard embedded computer and at more than 300 frames per second on a consumer laptop. We call our approach SVO (Semi-direct Visual Odometry) and release our implementation as open-source software. SVO: Fast Semi-Direct Monocular Visual OdometryChristian Forster, Matia Pizzoli, Davide Scaramuzza * Abstract-We propose a semi-direct monocular visual odometry algorithm that is precise, robust, and faster than current state-of-the-art methods. The semi-direct approach eliminates the need of costly feature extraction and robust matching techniques for motion estimation. Our algorithm operates directly on pixel intensities, which results in subpixel precision at high frame-rates. A probabilistic mapping method that explicitly models outlier measurements is used to estimate 3D points, which results in fewer outliers and more reliable points. Precise and high frame-rate motion estimation brings increased robustness in scenes of little, repetitive, and high-frequency texture. The algorithm is applied to micro-aerial-vehicle stateestimation in GPS-denied environments and runs at 55 frames per second on the onboard embedded computer and at more than 300 frames per second on a consumer laptop. We call our approach SVO (Semi-direct Visual Odometry) and release our implementation as open-source software.

Appearance-based Active, Monocular, Dense Reconstruction for Micro Aerial Vehicles

“…The derivation of the depth uncertainty reported in Equations (8) - (14) is similar to the one presented in [19], however with one critical difference that occurs in Equation (11). In the present paper, the disparity uncertainty σ p is a function of the appearance (i.e., texture) in the scene.…”

“…The inverse depthd u of a pixel u in the reference camera pose T w,r is a latent variable we infer from observations. An observation is a pair {I k , T w,k }, where we assume that T w,k is computed by an accurate visual odometry algorithm [11]. A measurement d u,k of pixel u is obtained by the k-th observation by triangulating from T r,k = T −1 w,r · T w,k and we assume it normally distributed with mean µ u,k and variance τ 2 u,k :…”

LSD-SLAM: Large-Scale Direct Monocular SLAM

“…In [9], a semi-dense depth filtering formulation was proposed which significantly reduces computational complexity, allowing real-time operation on a CPU and even on a modern smartphone [22]. By combining direct tracking with keypoints, [10] achieves high framerates even on embedded platforms. All these approaches however are pure visual odometries, they only locally track the motion of the camera and do not build a consistent, global map of the environment including loop-closures.…”

“…Two major reasons are (1) their use in robotics, in particular to navigate unmanned aerial vehicles (UAVs) [10,8,1], and (2) augmented and virtual reality applications slowly making their way into the mass-market.…”

Computationally efficient visual–inertial sensor fusion for Global Positioning System–denied navigation on a small quadrotor

“…2 In the past 5 years, many prominent research institutions began to develop advanced monocular visualbased simultaneous localisation and mapping (mSLAM) algorithms based on structure from motion (SFM) theory, [3][4][5][6][7][8][9][10][11] which are suitable to modern onboard embedded computers. Moreover, the visual scale problem, which was the main challenge of involving monocular vision into the control loop, has been addressed by fusing onboard inertial measurements (accelerometer and gyroscope), called the visualinertial navigation system (VINS).…”