MFA: A Smart Glove with Multimodal Intent Sensing Capability
Algorithm 2
Input: Fingertip pressure threshold , curvature value , pressure value vector group , curvature vector group , rotation angle and movement velocity generated by the posture sensor, action library
Output: Experimental intent probability set under sensor channel.
(1)
.
(2)
Vectorization of the pressure value information for each finger in , .
(3)
Vectorization of the curvature value information for each finger in The curvature dimension of the five-dimensional vector a is divided into [−180°, −90°), [−90°, −30°), [−30°, 30°), [30°, 90°), [90°, 180°), and when the curvature value of finger i is obtained, can be placed under the corresponding dimension of vector ..
(4)
Establishing the mapping between and ,.
(5)
Establishing the mapping of ,, and ,.
(6)
Using the Euclidean distance formula in multidimensional space to find the distance between the vector and the action vector in the action library ,.
(7)
Calculate the probability of the user performing action in the action library ,.
