Quantifying the Accuracy of a Motion Detecting Neural Prosthesis

Autor: Paul M. Yanik, Pablo Valenzuela, Premkumar Subbukutti, Martin L. Tanaka, David Hudson, Kimberly Hudson
Rok vydání: 2020
Předmět:
Zdroj: Volume 5: Biomedical and Biotechnology.
DOI: 10.1115/imece2020-23219
Popis: The neural prosthesis under development is designed to improve gait in people with muscle weakness. The strategy is to augment impaired or damaged neural connections between the brain and the muscles that control walking. This third-generation neural prosthesis contains triaxial inertial measurement units (IMUs - accelerometers, gyroscopes, and processing chip) to measure body segment position and force sensitive resistors placed under the feet to detect ground contact. A study was conducted to compare the accuracy of the neural prosthesis using a traditional camera motion capture system as a reference. The IMUs were found to accurately represent the amplitude of the gait cycle components and generally track the motion. However, there are some differences in phase, with the IMUs lagging the actual motion. Phase lagged by about 10 degrees in the ankle and by about 5 degrees in the knee. Error of the neural prosthesis varied over the gait cycle. The average error for the ankle, knee and hip were 6°, 8°, and 9°, respectively. Testing showed that the neural prosthesis was able to capture the general shape of the joint angle curves when compared to a commercial camera motion capture system. In the future, measures will be taken to reduce lag in the gyroscope and reduce jitter in the accelerometer so that data from both sensors can be combination to obtain more accurate readings.
Databáze: OpenAIRE