IDENTIFICATION OF THE INERTIA MATRIX OF THE UPPER LIMB

Abstract

INTRODUCTION Accurate results from biomechanical investigations are dependent of the body segment inertial parameters (BSIP) estimations. Actually, no routine technique provides personified BSIP for each subject studied. As a result, BSIP are commonly estimated using regression equations from literature [1]. Roboticians, who face similar problem with robots, use identification methods. Identification consists in solving a redundant system (i.e. the robot motion and its reactions on a bedplate are not dependent of the internal joint forces [2]) by numerical computations. This paper presents an application of the identification method for estimating the inertia matrix of the whole upper limb (upper arm, forearm, hand) modelled as a single rigid segment. Our results are compared with classical BSIP. METHODS Three subjects (S1, S2 and S3) performed specific movements (i.e flexion/extension, abduction/adduction and circumduction) with their right upper limbs. Subjects were modelled as being two segments: the trunk (S1) and the upper limb (S2). They were asked to keep the elbow straight and to move only the upper limb. Ten markers (S1: 6; S2: 4) were placed on the skin and collected by a motion analysis system (Saga3). Force platform data (Kistler) were synchronised. The glenohumeral joint centre (GHC) was computed using a functional method [3]. A matrix form is used to write the equations of motion. The inertia parameters of the 2 segments are contained in [H], a 7×7 positive definite matrix (the indices (1) and (2) refers to S1 and S2 respectively):