Congratulation to Raul Simpetru for finishing his Bachelor’s Thesis!
Congratulations to Raul Simpetru for completing his bachelor’s thesis on “Robust prediction of hand kinematics from surface electromyography using Convolutional Neural Networks and Machine Vision”.
In his thesis, Raul addressed the problem that most immersive human-computer interfaces available today are camera-based and can only provide insights into kinematics, not kinetics. He tested if neural interfaces can overcome these problems by extracting the output from the neural cells that generate movement, i.e., the motor units. Surface electromyography (sEMG) has the potential to replace digital cameras by extracting the final common neural code of movement transmitted from alpha-motoneurons in the spinal cord to groups of innervated muscle fibers. However, sEMG has not been shown to achieve many degrees of freedom comparable to the human hand while maintaining high predictive accuracy.
In his bachelor’s thesis, Raul proposes a new deep learning model that can learn all degrees of freedom (22) of the hand from sEMG data. His neural interface was able to learn the kinetics and kinematics over the entire force range of the digits of the hand in natural motion tasks.
His bachelor’s thesis shows that neural interfaces are the definitive immersive interface for human-computer interaction. Given its high performance, he supports that sEMG has the potential to revolutionize all industries that rely on computers, such as virtual reality, because grasping in these systems is based on contact with an object rather than force applied to the object, resulting in a disembodied experience. Since his system outperforms previous interfaces, we expect that amputees will most likely benefit from it since the output of our model can be adapted as control information for prosthetic limbs.
Congratulations again to Raul Simpetru for his impressive work! We are happy that he will continue his work in our lab and wish him good luck in his future research.