Mapping brain function and safer autonomous vehicles are the focus of the Schmidt Transformative Technology fund

Two projects – one that maps the function of the brain’s neuronal network in unprecedented detail and another that combines robotics and light-based computing circuits to create safe self-driving vehicles – have been funded through the Eric and Wendy Schmidt Transformative Technology Fund from Princeton.

The fund aims to stimulate momentum in science or engineering by creating entirely new technologies that can have a major positive impact on a research area. The fund supports projects that are so exploratory in nature that they are often considered too risky to qualify for financing from conventional sources such as government grants.

“Both projects bring together leading faculty from different areas of expertise to collaborate on efforts that, if successful, could dramatically advance what we know and what we can do,” said Peter Schiffer, dean of research at Princeton University, professor of physics. “The purpose of the Schmidt Fund is truly reflected in these examples of how new thinking and creative activity can drive transformational global impact.”

The fund was established in 2009 through a donation from Eric and Wendy Schmidt. Eric Schmidt is the co-founder of Schmidt Sciences, The Schmidt Family Foundation and Schmidt Ocean Institute, the former Chief Executive Officer of Google, and former Executive Chairman of Alphabet Inc., Google’s parent company. Wendy Schmidt is co-founder of Schmidt Sciences, and president and co-founder of The Schmidt Family Foundation and Schmidt Ocean Institute. Eric Schmidt received his bachelor’s degree in electrical engineering from Princeton in 1976 and served as a Princeton Trustee from 2004 to 2008.

The two newly funded technologies are:

Technology to track brain-wide signals via a new imaging system

• Mala Murthy, the Karol and Marnie Marcin ’96 Professor, professor of neuroscience and director of the Princeton Neuroscience Institute
• Andrew Leifer, assistant professor of physics and neuroscience

Mala Murthy (left) and Andrew Leifer (right) use the whole-brain map of a fruit fly (top) to track perturbations from one neuron across all 130,000 neurons.

This project will provide the most detailed insights yet into how signals propagate through networks of neurons to trigger brain functions. Based on extensive previous studies of neural connections in worms by Leifer’s team and fruit flies by Murthy’s team, the two scientists and their teams will develop a system to create maps of functional connectivity that will shed light on how brain activities such as controlling decision-making. and movement.

This new system allows researchers to discover how disrupting one neuron affects the signals traveling through neural pathways throughout the entire 130,000-neuron fruit fly brain. Current technologies would take more than a year to conduct such a large study, so the team will develop a system that simultaneously stimulates groups of neurons and images them using a new method called light bead microscopy, combined with machine learning to understand the decipher responses from the network. The technology will enable new discoveries in neuroscience and provide insight into how neural circuits give rise to behavior.

Revolutionizing robot safety through photonic computing

Self-driving cars and other robotic systems that rely on split-second decisions will be much safer when photon-driven computer chips developed by Paul Prucnal (left) are integrated into robotic systems developed by Jaime Fernández Fisac ​​(right).

This project aims to improve the safety and reliability of autonomous vehicles and robots using processors known as neuromorphic photonic chips, which work in a manner analogous to neurons and perform calculations using light, or photons , instead of electrons.

By combining photonic chips developed by Prucnal with robotic systems developed by Fisac, the researchers aim to build autonomous systems that can quickly evaluate safety-critical situations, such as a bicycle coming into the vehicle’s path, and choose the appropriate response, such as swerving. to avoid a collision.

Compared to regular electronic processors, the neuromorphic photonic chips process information much faster and can evaluate thousands of potential scenarios in real time, allowing the robot to choose the best emergency plan to avoid an accident. The approach will improve the safety of self-driving cars, delivery robots and other autonomous systems whose safe operation may depend on split-second decisions