Validation of Finger and Hand Movements Using the Xsens Metagloves and Kinfinity Gloves Inertial Motion Capture Gloves Against the Gold Standard of Optical Motion Capture

Status:

completed

Aims:

The aim of the study is to validate the Xsens Metagloves. To this end, the study will, on the one hand, validate standardised movements performed by the robotic arm and, on the other hand, record hand-related assessment methods to generate reference values for adults. This is based on the modification of an existing hand model.

Activities/Methods:

For the validation of the Manus Metagloves Pro, a laboratory study was first conducted at the Institute of Robotics and Mechatronics of the German Aerospace Center (DLR) in Munich using the robotic hand of the humanoid robot David. The aim was to assess how accurately the Manus Metagloves Pro could capture predefined finger movements of the robotic hand. To this end, the correlation between the directly programmed and internally measured joint angles of the robotic hand and the indirectly measured joint angles obtained from the Manus Metagloves Pro was analyzed. In a second study involving human participants, the Manus Metagloves Pro were compared with the Vicon Optical Motion Capture System. In this case, the correlation between two systems based on the indirect measurement of joint angles was investigated. Thus, in addition to the internal validation performed with the robotic hand, the second study provided a direct comparison between two independent measurement systems.

Results:

The comparison between the Manus Metagloves Pro and the humanoid robot David showed a highly variable agreement of joint angles depending on the finger, joint, and type of movement. For the index finger MCP joint, high correlations were observed during flexion, whereas abduction movements showed considerably lower values. PIP joints were in the moderate range, while DIP joints were generally more weakly correlated, although higher values were also observed during isolated movements. Thumb movements showed overall very low correlations. Overall, the temporal patterns of both systems were in good agreement, but the absolute joint angle values differed substantially.

In the Manus vs. Vicon comparison, similarly strong differences were observed between joints and movement directions. MCP and PIP joints of the long fingers generally showed moderate to high correlations, whereas DIP joints were considerably lower. The thumb showed a mixed pattern with very low to moderate values. The ICC values were inconsistent, being high in some cases (e.g., thumb MCP) but low for many long-finger joints. For abduction-adduction, overall correlations were low, while ICC values ranged from moderate to high. In the 3D marker analysis, the highest correlations were found in the distal finger regions, particularly in the dorsopalmar direction, whereas radial-ulnar showed substantially weaker relationships. Depending on the task, the best agreement was observed during dynamic flexion movements, especially at the PIP joints, while static conditions and isolated movements showed considerably lower or no correlation.

Last Update: