Optical Flow Aided Motion Estimation for Legged Locomotion

Abstract

Abstract-Dynamic legged locomotion entails navigating terrain at high speed. The impact shocks from rapid footfalls, pivotal for such mobility, introduce large impulses that saturate motion measurement. A biomimetic approach is presented in which visual information, in the form of optical flow, complements information from inertial sensors. The motion is then determined using a two-phase Hybrid Extended Kalman Filter. Experimentation in determining attitudes on a robotic leg platform shows a reduction in drift over inertial approaches and in delay over visual approaches. In tests with 6g impulses, pose was recovered within 5 deg rms with angular rate errors limited to 10 deg/sec at frequencies up to 250 Hz.

A Hybrid Motion Model for Aiding State Estimation in Dynamic Quadrupedal Locomotion

“…In previous work [5], the use of optical flow as a lowfrequency complementary measure for aiding high-frequency inertial measurements was explored and found to estimate orientation as long as it was sufficiently initialized. Figure 3 illustrates the integrative approach that is used to limit drift.…”

“…Compact, self-contained sensing with respect to a bodycentered inertial frame is typically achieved using an inertial measurement unit (IMU) [5]. In the legged domain, this is complicated by footfall shocks and sensor misalignment leading to errors in the compensation of gravitational acceleration.…”

Object motion detection using information theoretic spatio-temporal saliency

“…This approach approximates the object motion by estimating vectors originating or terminating at pixels in image sequences, so it represents the velocity field which warps one image into another high dimensional feature space. Some researchers [12,13] proposed motion detection methods based on optical flow technique, these methods can accurately detect motion in the direction of intensity gradient, but the motion which is tangential to the intensity gradient cannot be well represented by the feature map. Moreover, optical flow based methods also suffer from the illumination problem.…”

“…Many methods have been developed and reported in the last two decades for object detection from video. These methods can be categorized in three approaches, namely contour-based [9][10][11], orientation-based [12,13] and distribution-based [2,[14][15][16][17][18][19]. The contour-based approach is able to give a good localization of object contour, but it may not be able to handle fast motion.…”

Heading direction anticipation using visual feedback system

“…Study regards on visual odometry has attracted many researchers such as Kazik and Goktogan that suggested the visual odometry method to estimate the location and orientation based on Fourier-Merlin transform and phase-only matched filters [6]. Other works related includes in mobile robots [7][8][9][10], unmanned air vehicle (UAV) [11][12], automotive [13][14] and humanoid robots [15][16]. However, there have some assumptions and limitations included with it in order to successfully implement it, such as [6,7].…”