Recognition of ascending stairs from 2D images for control of powered lower limb prostheses

Intent recognition is essential for effective control of powered assistive devices, such as powered lower limb prostheses, exoskeletons, or wheelchairs. Currently, EMG and mechanical sensors are used for intent recognition of powered lower limb prostheses. We propose the addition of vision for improved intent recognition control, with this work focused on determining the best method for recognizing of ascending stair edges from 2D images. In this work different image processing methods were tested to determine which method produces the best line extraction. The best results were obtained using Canny, Sobel, Prewitt, and Roberts Cross edge detectors for four colorspace components. Finally, a convex/concave line decision system was developed to produce preliminary results about the presence or absence of stairs in a given image.

[1]  John F. Canny,et al.  A Computational Approach to Edge Detection , 1986, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[2]  Fan Zhang,et al.  Continuous Locomotion-Mode Identification for Prosthetic Legs Based on Neuromuscular–Mechanical Fusion , 2011, IEEE Transactions on Biomedical Engineering.

[3]  N. Ranganathan,et al.  Gabor filter-based edge detection , 1992, Pattern Recognit..

[4]  Michael Brady,et al.  Vision-based Detection of Stair-cases , 2003 .

[5]  刘霁,et al.  A stairway detection algorithm based on vision for UGV stair climbing , 2008 .

[6]  Robert D. Lipschutz,et al.  Robotic leg control with EMG decoding in an amputee with nerve transfers. , 2013, The New England journal of medicine.