Vision-only egomotion estimation in 6DOF using a sky compass

Abstract

SUMMARYA novel pure-vision egomotion estimation algorithm is presented, with extensions to Unmanned Aerial Systems (UAS) navigation through visual odometry. Our proposed method computes egomotion in two stages using panoramic images segmented into sky and ground regions. Rotations (in 3DOF) are estimated by using a customised algorithm to measure the motion of the sky image, which is affected only by the rotation of the aircraft, and not by its translation. The rotation estimate is then used to derotate the optic flow field generated by the ground, from which the translation of the aircraft (in 3DOF) is estimated by another customised, iterative algorithm. Segmentation of the rotation and translation estimations allows for a partial relaxation of the planar ground assumption, inherently increasing the robustness of the approach. The translation vectors are scaled using stereo-based height to compute the current UAS position through path integration for closed-loop navigation. Outdoor field tests of our approach in a small quadrotor UAS suggest that the technique is comparable to the performance of existing state-of-the-art vision-based navigation algorithms, whilst also removing all dependence on additional sensors, such as an IMU or global positioning system (GPS).

An ant-inspired celestial compass applied to autonomous outdoor robot navigation

“…This finding will have to be confirmed in further experiments including outbound trajectories with various shapes and distances. At last, the relative mean homing error of 0.7% performed with AntBot is currently rather good in comparison with the current state-of-the-art in visual odometry, such as the 1.3% relative precision of a sky compass-based egomotion estimation system [73]. The different ways in which celestial cues are acquired could also be one of the reasons why the relative error was lower in the case of Sahabot 2 than AntBot.…”

An integrated position estimation by using optical flow for visual robot navigation

“…This OF method is to precisely estimate ∆Mt from the images ft-1 and ft. To do this, the ft-1 is shifted by two reference amounts ±∆xref along x axis to get two translation reference images fx±, and is shifted by two reference amounts ±∆yref along y axis to get two translation reference images fy±, and is rotated by two reference amounts ±∆rref about z axis to get two rotation reference images fr± respectively. From [7], the image-interpolation algorithm assumes that the image deformation would be continuous and linear along with camera motion. Hence, the image ft can be approximated by…”

“…To estimate visual robot's motion and position, a honeybee-vision inspired OF measurement method is introduced into our relative position estimation that estimates global image motion with a simple and non-iterative implementation [7]. We consider a mobile robot translating along two axes (x, y) and rotating about the third axis (z).…”

A Computer Vision Milky Way Compass

“…There have been past insect biology-inspired studies using celestial cues for navigation. The sky itself can be a reference in day time with sun, clouds and scattering [20], resulting in patterns that are stable [21]. Some implementations have been developed by using sky-polarised light [20] to achieve autonomous navigation both on the ground [22,23] and as part of a flying navigation system on a drone [24].…”