DocumentCode :
2628942
Title :
INS/EKF-based stride length, height and direction intent detection for walking assistance robots
Author :
Brescianini, Dario ; Jung, Jun-Young ; Jang, In-Hun ; Park, Hyun Sub ; Riener, Robert
Author_Institution :
Dept. of Mech. & Process Eng., ETH Zurich, Zurich, Switzerland
fYear :
2011
fDate :
June 29 2011-July 1 2011
Firstpage :
1
Lastpage :
5
Abstract :
We propose an algorithm used to obtain the information on stride length, height difference, and direction based on user´s intent during walking. For exoskeleton robots used to assist paraplegic patients´ walking, this information is used to generate gait patterns by themselves in on-line. To obtain this information, we attach an inertial measurement unit(IMU) on crutches and apply an extended kalman filter-based error correction method to reduce the phenomena of drift due to bias of the IMU. The proposed method is verifed in real walking scenarios including walking, climbing up-stairs, and changing direction of walking with normal.
Keywords :
Kalman filters; diseases; error correction; gait analysis; inertial navigation; medical disorders; medical robotics; medical signal processing; patient rehabilitation; spatial variables measurement; EKF based error correction method; IMU; INS-EKF; crutches; direction intent detection; ensemble Kalman filter; exoskeleton robots; gait pattern generation; height detection; inertial measurement unit; inertial navigation system; paraplegic patients; stair climbing; stride length detection; user intent; walking assistance robots; walking direction change; Exoskeletons; Legged locomotion; Measurement uncertainty; Navigation; Robot sensing systems; Trajectory; Gait; Humans; Models, Theoretical; Robotics; Self-Help Devices; Walking;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Rehabilitation Robotics (ICORR), 2011 IEEE International Conference on
Conference_Location :
Zurich
ISSN :
1945-7898
Print_ISBN :
978-1-4244-9863-5
Electronic_ISBN :
1945-7898
Type :
conf
DOI :
10.1109/ICORR.2011.5975363
Filename :
5975363
Link To Document :
بازگشت