Machine learning for printed electronics
- Lassonde Undergraduate Research Award- summer research
- NSERC USRA
Position Title: Research Assistant/summer researcher
Location: Bergeron Center, York University
Professor: Gerd Grau
Department: Electrical Engineering and Computer Science
Contact for Professor: email@example.com, 416-736-2100 x70127
# of positions available: 1
Title: Machine learning for printed electronics
Printed electronics is an emerging technology to fabricate microelectronic devices. The use of printing techniques offers several advantages over traditional microfabrication technology, which is used to fabricate systems such as silicon chips and displays today. These traditional methods achieve very high performance, however, at a high cost. Printing allows microelectronics to be fabricated at significantly lower cost especially for large-area systems due to high fabrication speeds and because printing is a type of additive manufacturing. In addition, printed electronics can be fabricated on novel substrates such as plastic or paper. This enables new applications such as bendable displays, low-cost RFID tags or large, low-cost sensor networks.
Whilst this new technology shows great promise, there are still a number of challenges making printed electronics a very active area of research. There is one position available for an undergraduate researcher this summer working on the following project:
A very important aspect of printed electronics is the design of patterns that are digitally input into the printer to control the deposition of ink onto the substrate. This needs to consider non-idealities that occur because liquid ink can flow after being deposited on the substrate especially in micron-scale patterns. Work has been done by this group and others to analyze this flow, which very quickly becomes very complex especially for large patterns. It is therefore difficult to design large circuits. Currently, either manual tweaking is used or the printed patterns are non-optimal for example with lower resolution than theoretically possible. This project aims to overcome this limitation by machine learning. The student will create a machine learning algorithm that takes in an arbitrary circuit design and returns the printer commands that will print this circuit with as little error as possible. The algorithm will be trained on real data obtained with a micro-inkjet printer in the lab. The student will train the algorithm over multiple generations through subsequent prints to optimize its performance.
Duties and Responsibilities of the student:
- Program machine learning algorithm
- Plan, carry out and analyze experiments
- Learn the required techniques and skills under supervision
- Follow safe lab procedures
- Read relevant literature
- Present research results in regular meetings
- Write up results to be submitted for publication or to be part of a future publication
Skills and Qualifications:
- The student needs to have programming experience as well as some background in machine learning.
- Experimental lab skills would be beneficial but can be learned during the summer.
- The student needs to be able to work independently as well as effectively in a team of researchers. He or she needs to be enthusiastic to learn new knowledge and apply it to challenging research questions.
Degrees, courses and Disciplines prerequisite*: Engineering, Computer Science or Physical Sciences
Open to external students: No
Duration: 16 weeks minimum
Start Date: 05/01/2018 (estimated)
End Date: 08/31/2018 (estimated)
If you are interested in this research project, please contact Dr. Gerd Grau, at firstname.lastname@example.org, 416-736-2100 x70127