Investigate Electroencephalogram in Human-Robot Collaboration and Combat Team
Abstract
Human-robot collaboration and combat teaming is an emerging technology of defense. It not only provides high performance computational power for human in the dynamic environment, but also coordinates both human and robots for performing efficient actions. Although robots can be completely controlled by human hands via remote controllers, human hands cannot be released for other important actions. Some other media like voice controller can release hands to instruct and control robots, however, dynamic environments with noise or inaccurate speech recognition may cause inefficient control over robots in real-time dynamic environment. The project will propose a new instrumental platform for students and faculty in robotics research, teaching, and education at University of Houston-Downtown (UHD). Specifically, we propose to invest a new instrumental equipment platform for investigating Electroencephalogram (EEG) signals via Brain Computer Interface (BCI) to control mobile robots in human-robot collaboration teaming and educating underrepresented minority students on robotics study in Houston metropolitan area. Bioelectrical signals of human brain will be acquired, measured, and analyzed as the mind-controller to interact with robots. However, human brain signal is usually nonstationary signal which has different distribution on different human subjects. Even if for the same subject, signal distribution of the same intention and sensation data is different when subjectÕs brain signal is acquired at different time slots. In addition, it is impossible to collect and label all types of EEG data from all subjects. Therefore, EEG records are limited by scarcity of meaningful data labels. Finally, consistent intention and sensation are important between human-robot collaboration. Rapid changes of dynamic environment require real-time rapid learning models for classifying EEG data based intention and sensation signals. The proposed research project related to human-robot collaboration and combat teaming focuses on interdisciplinary research on robotics, neuroscience, and machine learning. Robustness of BCI based robotics control should be improved for generating stable control signals. Nonstationary signals are investigated by transfer learning methods for enhancing robustness and stability of brain-controlled robotsÕ actions. For scarcity of EEG data labels, data augmentation techniques will be investigated based on limited labeled data for increasing amount of labeled data. Online machine learning models are also investigated for dynamically identifying human intention and sensation EEG signals in a real-time mode. A closed-loop system will be investigated by interleaving subject feedback, data collection, and model training. If the proposed methods are successful, they are able to improve robustness, efficiency, speed, and accuracy of BCI based human-robot interaction. As a minority and Hispanic-Serving Institution (HSI), University of Houston-Downtown is a comprehensive four-year university offering Bachelor s and a limited number of Master s degrees and providing strong academic and career preparation as well as life-long learning opportunities. UHD is also an inclusive community dedicated to integrating teaching, service, and scholarly research to develop students talents with 21st century skills and prepare them for success in a dynamic global society. Both hardware and software demands of new robotic applications require the cutting-edge robotics resources and facilities. The project will provide minority and underrepresented students the most exciting robotics projects and cutting-edge techniques.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Oct 23, 2018
- Source ID
- W911NF1710182
Entities
People
- Yuchou Chang
Organizations
- Army Contracting Command
- United States Army
- University of Houston–Downtown