Egomotion Recovery Research Articles

Towards Privacy-Preserving Ego-Motion Estimation Using an Extremely Low-Resolution Camera

Ego-motion estimation is a core task in robotic systems as well as in augmented and virtual reality applications. It is often solved using visual-inertial odometry, which involves using one or more always-on cameras on mobile robots and wearable devices. As consumers increasingly use such devices in their homes and workplaces, which are filled with sensitive details, the role of privacy in such camera-based approaches is of ever increasing importance. In this letter, we introduce the first solution to perform privacy-preserving ego-motion estimation. We recover camera ego-motion from an extremely low-resolution monocular camera by estimating dense optical flow at a higher spatial resolution (i.e., 4x super resolution). We propose SRFNet for directly estimating Super-Resolved Flow, a novel convolutional neural network model that is trained in a supervised setting using ground-truth optical flow. We also present a weakly supervised approach for training a variant of SRFNet on real videos where ground truth flow is unavailable. On image pairs with known relative camera orientations, we use SRFNet to predict the auto-epipolar flow that arises from pure camera translation, from which we robustly estimate the camera translation direction. We evaluate our super-resolved optical flow estimates and camera translation direction estimates on the Sintel and KITTI odometry datasets, where our methods outperform several baselines. Our results indicate that robust ego-motion recovery from extremely low-resolution images can be viable when camera orientations and metric scale is recovered from inertial sensors and fused with the estimated translations.

Efficient tracking and ego-motion recovery using gait analysis

We present a strategy based on human gait to achieve efficient tracking, recovery of ego-motion and 3-D reconstruction from an image sequence acquired by a single camera attached to a pedestrian. In the first phase, the parameters of the human gait are established by a classical frame-by-frame analysis, using an generalized least squares (GLS) technique. The gait model is non-linear, represented by a truncated Fourier series. In the second phase, this gait model is employed within a “predict–correct” framework using a maximum a posteriori, expectation-maximization (MAP-EM) strategy to obtain robust estimates of the ego-motion and scene structure, while continuously refining the gait model. Experiments on synthetic and real image sequences show that the use of the gait model results in more efficient tracking. This is demonstrated by improved matching and retention of features, and a reduction in execution time, when processing video sequences.

Robotic sensing for the partially sighted

Partial sightedness is a sensory disability which can to some extent be alleviated by artificial aids. Many of the sensory methods used in robotics can be applied in attempts to recapture some of the sensory information a partially sighted person has lost. This paper describes a device which uses sonar and stereo vision sensors for this task. The device is portable, and is worn by the user, giving them freedom of movement over kerbs, stairs and rough ground. Sensor motion during walking is measured using visual egomotion recovery and odometry, and has been modelled to allow compensation in the sensor readings. A ground position estimate is continually updated by scene ground-plane fitting, or from the walk-motion model, and is used to classify scene features as obstacles or parts of the ground. Methods for the robust reconstruction of image points and lines into scene features are developed. The recognition of world objects of exceptional significance to a mobile person - kerbs and stairs - is given particular attention. A user interface, which has undergone limited real world testing, is also described. Experimental results are presented for the various parts of the system.

Robust estimation of egomotion from normal flow

This paper investigates the robustness of an algorithm due to Horn and Weldon (1988) for recovery of egomotion from optic-flow. Assuming only normal components of flow vectors are availableand that 3D angular velocity is known, tight constraints can be constructed on the direction of translational motion or, equivalently, on the focus of expansion. In practice however this is unacceptably restrictive. Some allowance must be made for uncertainty in angular velocity. We show that the algorithm can indeed be extended to cope with such uncertainty with graceful degradation in accuracy of estimated position of the focus of expansion. The shape of the error distribution depends on whether the focus of expansion is inside or outside the field of view. If it is inside, the error distribution is isotropic. As it moves outside the distribution becomes increasingly anisotropic. Results from an implementation of the algorithm confirm the validity of the error bounds.

The role of fixation in visual motion analysis

How does the ability of humans and primates to fixate at environmental points in the presence of relative motion help their visual systems in solving various tasks? To state the question in a more formal setting, we investigate in this article the following problem: Suppose that we have an active vision system, that is, a camera resting on a platform and being controlled through motors by a computer that has access to the images sensed by the camera in real time. The platform can move freely in the environment. If this machine can fixate on targets being in relative motion with it, can it solve visual tasks in an efficient and robust manner? By restricting our attention to a set of navigational tasks, we find that such an active observer can solve the problems of 3-D motion estimation, egomotion recovery, and estimation of time-to-contact in a very efficient manner, using as input the spatiotemporal derivatives of the image-intensity function (or normal flow). Fixation over time changes the input (motion field) in a controlled way and from this change additional information is derived making the previously mentioned tasks easier to solve.