Inertial navigation algorithm for trajectory of front-wheel walker estimation

Abstract

In this paper, we propose a system for trajectory of walker estimation. The system consists of an inertial measurement unit (IMU) and two encoders attached to a front-wheel walker. The IMU is employed to estimate the trajectory of the walker while the encoders are used to update the trajectory of the walker during rolling on the floor. Three update equations are proposed: quaternion update using the vertical vector, quaternion update using the yaw angle of the walker and position update using encoders. We implemented an experiment which focused on four walking styles of: continuous rolling, step by step rolling, complete lifting and 2 back tips lifting. Results of the experiment show the appropriateness of proposed update equations in all cases in general and in continuous rolling in particular.

Fast calibration for parameters of an inertial measurement unit fixed to a standard walker

“…In comparison with other research studies, the results of [ 10 ] and [ 19 ] were used, in which the translation vector is manually and carefully measured. Because there is no rolling movement for the standard walker, the results of the lifting case in [ 10 ] and [ 19 ] were used in the comparison. In the lifting case, the accuracy of distance was estimated at approximately 98.3% in [ 10 ], 98.5% in [ 19 ], and 97% in this study.…”

“…At the beginning of the period, the -axis and -axis of the BCS are on the horizontal floor, while the -axis points upward, as shown in Figure 2 . Unlike the procedures followed in [ 10 ] and [ 19 ], the origin is set at the midpoint of segment , as illustrated in Figure 1 , because the distance from the origin to the heels of a user is nearest after a moving step. The -axis of the WCS points upward, while the -axis is in the vertical plane with the -axis.…”

“…The trajectory of the IMU is estimated using the basic KF based on the INS, which is not a new result, and can be found in [ 19 , 21 , 22 ]. The trajectory consists of position , quaternion , and velocity of the IMU in the WCS.…”

“…Since the velocity and the height of the walker are almost zero, the error in the velocity and the third component of the walker's position can be reset. The algorithm to detect the zero velocity intervals and the updating equations for the KF at the intervals can be found in [ 19 ].…”

“…The term is the translation vector computed by (15) using the full procedure consisting of four tilting movements, while is computed by (19) using the fast procedure consisting of two tilting movements. Since it is difficult to evaluate the accuracy of parameters in Table 1 , and the main purpose of the proposed system is to estimate the walking parameters using the trajectory of the walker, we implemented two experiments to evaluate the accuracy of the proposed system by the estimated distance and position, in which the trajectory of the walker (BCS) is computed from the trajectory of the IMU using the rotation matrix and translation vector [ 19 ]: where is the position of the BCS; and is the rotation matrix from the BCS to the WCS at discrete-time index .…”