Y-Prize 2021 – 2022: Nanostructured Membranes and Semantic SLAM

The Y-Prize is an annual competition with the mission of inventing new products using technology being developed at Penn. The challenge brings together Engineering, Wharton and the Penn Center for Innovation and has produced some notable inventions since its launch at Penn in 2012.

Penn’s Y-Prize was designed as the inverse of the X-Prize, a competition which poses real-world problems, asking participants to find the best technological solution. The Y-Prize is unique in that it identifies the technology first, and then asks students to imagine a real-world problem that the technology can solve. With the technology already being developed, products that come out of the Y-Prize have the potential to be used immediately.

To encourage that quick translation, the winning team is awarded $10,000 to take their idea from theory to practice. Any Penn student can participate in the challenge, with applications open until March 7th.

Last year’s winning team was “Metal Light,” which proposed using metal-air scavenger technology developed in the lab of James Pikul, assistant professor in the Department of Mechanical Engineering and Applied Mechanics. Their sustainable light works off the grid; when attached to a piece of scrap metal, it extracts energy from an oxidation reaction fueled by the surrounding air.

The Metal Light was not the only winning product. Coming in second was a team that developed shipping container sensors with the same metal-air scavenging technology that could detect damage, theft and human trafficking.

This year new technologies are “Nanostructured Membranes” from Chinedum Osuji, Eduardo D. Glandt Presidential Professor in the Department of Chemical and Biomolecular Engineering, and “Semantic SLAM” from researchers from the GRASP Lab .


Nanostructured Membranes     

Membrane technology is important for water filtration, chemical separation, and drug delivery. However, current membrane technology is faced with the trade-off of permeability and selectivity, meaning that a membrane has to sacrifice purity for speed or vice versa. New nanostructured membrane technology allows for high selectivity without compromising permeability, which allows this membrane to produce pure water quickly and cheaply.

The holes that allow for liquid to pass through a membrane are usually different sizes, however, this new technology uses self-assembling crosslinks to create uniform, nanoscale pores. Ultraviolet light creates covalent bonds between polymers in the membrane, aligning them in a perfect chemical structure. The membrane itself is a thousand times thinner than a strand of hair and is capable of excluding very small molecules.

Many applications can be imagined with this technology, with some future ideas including use in industrial chemical separation and protection against chemical warfare agents. Read more here.


Semantic SLAM

Semantic SLAM is the introduction of vocabulary to robots for improved machine mapping. SLAM stands for “simultaneous localization and mapping,” a capability that autonomous machines have been using to navigate new environments for some time. However, the maps robots make with SLAM are not always easy for a human to understand. Semantic SLAM aims to improve human-robot communication by teaching machines recognize and label objects, landmarks, and places while making maps of a physical space.

For example, take a robotic vacuum, which moves around a home and makes a map for the objects it bumps into. However, the vacuum does not know if a given object is the leg of a chair or the edge of a kitchen island. Realizing the difference could mean the vacuum could move around the legs of the chair to clean underneath it. This might be a minor improvement in keeping a house clean, but in the context of autonomous vehicles, this kind of semantic identification has major safety implications.

This learning ability is done through a machine learning process called data association. When the robot encounters an object, it will hypothesize many possible identifications and responses. Data association matches those hypotheses to a stored database of objects, places, and landmarks to allow the robot to choose its next move. Semantic SLAM has many applications such as inventory management and agricultural crop tracking and counting. The GRASP lab is hoping that students will use fresh eyes and wide perspectives to create inventions based on this technology. Read more here.