Date of Award


Degree Name

Master of Science


Electrical Engineering


Thompson Sarkodie-Gyan


This research study proposes a multi-sensor data fusion technique to determine the complex interactions between the sensory, muscular and mechanical components of the human locomotor systems (neuromechanics). The object in this work is to demonstrate the viability in using inertial sensors for accurate gait phase determination and to present the acceleration effect specific body segment contributes in order to establish a functional gait. The current method used in determining gait phases consist of combining ground reaction force and angular data, this method is time consuming and cannot be totally reliable. We found out that gait phases can be determined distinctively by measuring vertical and horizontal accelerations (X-Y) in the sagittal plane on the foot. The foot acceleration graph show 98% unique consistency in determining the seven gait phases, during the experiment when the subjects were walking at normal; fast; even running. Note that this foot acceleration pattern was accurate for three consecutive experiments carried out on three different days under the same experimental conditions. Although the shank, thigh and hip acceleration showed uniqueness in determining the gait phases when the subjects were walking at normal speed, we discovered irregularities in the graph pattern when the subjects were running.

For 132lbs, 5.6ft male subject walking at 0.9m/s we noticed that the highest negative acceleration (-0.460g, foot y-direction in the sagittal plane) occurred at heel strike and this is due to the slowing down of the muscles as the leg stabilizes from the swing phase of the sagittal plane. The corresponding vertical acceleration was 1.226g with a vertical ground reaction force of 24N. As the right leg transitions from heel strike to toe off, a maximum ground reaction force of 619N was detected. At toe-off maximum positive acceleration in the foot y-direction was recorded (0.327g) corresponding to the second highest peek of the vertical acceleration (x-direction 0.311g). We also found out that the only lowest negative acceleration in the vertical plane occurred at the cross over (initial swing) as the foot leaves the ground and the hip flexor muscles are activated to accelerate the leg forward.

In total six able bodied subjects were involved in this research (four male and two female). The experiment was performed using rate gyroscopes and linear accelerometers attached to the right hip, right thigh, right knee, right shank, right ankle and right foot. The assumption in this work is that walking pattern in able-bodied people is symmetrical, thus we assume the same acceleration conditions and gait phases detections as well is symmetrical. The subjects were allowed to walked at their normal walking speeds, however, for comparison in some cases all subjects walked at 0.9m/s and 1.08m/s. The experiment lasted for 120s when the subjects walked on an instrumented treadmill and the setup was synchronized with ground reaction force sensor, and Simi motion capture system. The sampling frequency was 280Hz for all four sensing technologies. When the subjects walked over ground, the experiments lasted for about 5sec. Outputs from the sensors were fed into fuzzy inference system, where the concept of fuzzy similarity was applied to determined the coordinated walking pattern for each subject (CWP).




Received from ProQuest

File Size

81 pages

File Format


Rights Holder

John Kweku Avor