Ambient motion estimation in dynamic scenes using wearable visual-inertial sensors

He, Hongsheng; Tan, Jindong

doi:10.1109/icra.2014.6906976

Cited by 4 publications

(5 citation statements)

References 14 publications

(19 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This will be beneficial for improving the performance in recognizing gestures which have similar trajectories but different hand shapes. We need to study motion estimation in dynamic 3D scene [51,52], which is to enable the robot to recognize gestures in motion. We also plan to design a real-time HRI system for practical application.…”

Section: Discussionmentioning

confidence: 99%

Adaptive Local Spatiotemporal Features from RGB-D Data for One-Shot Learning Gesture Recognition

Lin

Ruan

et al. 2016

Sensors

View full text Add to dashboard Cite

Noise and constant empirical motion constraints affect the extraction of distinctive spatiotemporal features from one or a few samples per gesture class. To tackle these problems, an adaptive local spatiotemporal feature (ALSTF) using fused RGB-D data is proposed. First, motion regions of interest (MRoIs) are adaptively extracted using grayscale and depth velocity variance information to greatly reduce the impact of noise. Then, corners are used as keypoints if their depth, and velocities of grayscale and of depth meet several adaptive local constraints in each MRoI. With further filtering of noise, an accurate and sufficient number of keypoints is obtained within the desired moving body parts (MBPs). Finally, four kinds of multiple descriptors are calculated and combined in extended gradient and motion spaces to represent the appearance and motion features of gestures. The experimental results on the ChaLearn gesture, CAD-60 and MSRDailyActivity3D datasets demonstrate that the proposed feature achieves higher performance compared with published state-of-the-art approaches under the one-shot learning setting and comparable accuracy under the leave-one-out cross validation.

show abstract

Section: Discussionmentioning

confidence: 99%

Adaptive Local Spatiotemporal Features from RGB-D Data for One-Shot Learning Gesture Recognition

Lin

Ruan

et al. 2016

Sensors

View full text Add to dashboard Cite

show abstract

“…In this paper, we predict the corresponding features based on attitude estimations using onboard IMU. The movement of interested features is predicted as (19) where is the pixel position of a feature, and is the movement of features caused by camera rotation [2] (20)…”

Section: A Active Feature Searchmentioning

confidence: 99%

“…Similarly, a motion hazard alarm is given when an object is running toward the blind. Relative motion is analyzed by fusing video input and IMU measurements [2].…”

Section: A System Setupmentioning

confidence: 99%

“…To specifically address missing necessary functions, we have been developing a wearable navigator on Google Glass to improve the safety and mobility of visually impaired and blind people [2]- [4]. Motivated by the vestibular and vision system in humans and animals, the blind navigation system provides the functionality of localization and scene perception by fusing visual and inertial sensors because of their complementary properties.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Wearable Ego-Motion Tracking for Blind Navigation in Indoor Environments

Guan

et al. 2015

IEEE Trans. Automat. Sci. Eng.

Self Cite

View full text Add to dashboard Cite

This paper proposes an ego-motion tracking method that utilizes visual-inertial sensors for wearable blind navigation. The unique challenge of wearable motion tracking is to cope with arbitrary body motion and complex environmental dynamics. We introduce a visual sanity check to select accurate visual estimations by comparing visually estimated rotation with measured rotation by a gyroscope. The movement trajectory is recovered through adaptive fusion of visual estimations and inertial measurements, where the visual estimation outputs motion transformation between consecutive image captures, and inertial sensors measure translational acceleration and angular velocities. The frame rates of visual and inertial sensors are different, and vary with respect to time owning to visual sanity checks. We hence employ a multirate extended Kalman filter (EKF) to fuse visual and inertial estimations. The proposed method was tested in different indoor environments, and the results show its effectiveness and accuracy in ego-motion tracking. Note to Practitioners-This paper was motivated by the necessary mobility needs of visually impaired and blind people.According to a review, the top listed accidents they experienced were head-level collision and motion collision, even though they were using assistive devices. Currently, assistive devices are available to emulate missed functions due to vision loss, such as text reading, object recognition, and knee-level obstacle avoidance (e.g., smart canes), but they do not provide the solution to the fundamental problems as raised in the review. The research team is developing a wearable blind navigator based on Google Glass. The target function modules include traversable region detection, motion estimation, and scene dynamics analysis. The ego-motion estimation method presented in this paper is part of the motion estimation module. The experiments on real scenarios suggest that this approach is feasible and effective. In future research, we will address the problems of robustness and real-time performance in system implementation.

show abstract

“…Similarly, a motion hazard alarm is given when an object moving toward the wearer. Relative motion is analyzed by fusing video inputs and IMU measurements [12]. The other supporting interactive functions are implemented as services on call.…”

Section: Wearable Blind Navigatormentioning

confidence: 99%

Demonstration Paper

Tan

2014

Proceedings of the Wireless Health 2014 on National Institutes of Health

Self Cite

View full text Add to dashboard Cite

This paper presents an ego-motion tracking method using visual-inertial sensors to assist the visually impaired and blind (VIB) people to travel in unknown dynamic environments. We focus on the ego-motion tracking functionality to inform the wearers of their relative position with respect to the environment. A traveled trajectory is recovered by concatenating the transformation estimated from visual correspondences and instantaneous movements captured by inertial sensors. Therefore, we introduce an adaptive mechanism to judge the reliability of visual tracking by comparing the estimated rotation with the gyroscopic measurement. The measuring frequencies of visual and inertial sensors are different because of different physical sampling rates and the introduced adaptive mechanism. We adopt the multi-rate extended Kalman filter (EKF) to fuse the visual estimation and inertial measurement. In the experiment, we wear the navigation system to follow a path in an indoor environment, and the results show the effectiveness and precision of the proposed methods in ego-motion tracking.

show abstract

Ambient motion estimation in dynamic scenes using wearable visual-inertial sensors

Cited by 4 publications

References 14 publications

Adaptive Local Spatiotemporal Features from RGB-D Data for One-Shot Learning Gesture Recognition

Adaptive Local Spatiotemporal Features from RGB-D Data for One-Shot Learning Gesture Recognition

Wearable Ego-Motion Tracking for Blind Navigation in Indoor Environments

Demonstration Paper

Contact Info

Product

Resources

About