By making key streetscape technologies available through open source channels, we enable cities and communities to adopt and improve upon our innovations. Our publicly available resources make it possible for organizations to implement the next generation of smart street solutions. Through public repositories and comprehensive documentation, CS3 supports a collaborative ecosystem where developers and researchers can examine, enhance, and build upon shared tools.
Thrust 1
Massive MIMO Direction-of-Arrival
This project develops a faster, more accurate way to determine the location of wireless devices in urban environments using advanced antenna systems. By improving location accuracy with just a single measurement, this technology could help emergency responders find people more quickly and enable better navigation services in smart cities.
MobLoc: CSI-based Location Fingerprinting with MUSIC
MobLoc is a new indoor positioning system that helps track the location of mobile devices with high accuracy using WiFi signals. By analyzing how WiFi signals bounce and travel through indoor spaces, MobLoc can determine device locations within about 1-3 feet - making it useful for applications like indoor navigation.
Thrust 2
Generative Camera Dolly
This project transforms regular street videos into dynamic multi-angle views, similar to having multiple cameras recording the same scene. By generating new viewpoints from a single video feed, it helps urban planners and safety researchers analyze streetscapes more comprehensively without needing additional physical cameras.
TeD-SPAD: Temporal Distinctiveness for Self-supervised Privacy-preservation for video Anomaly Detection
TeD-SPAD is a privacy-focused video monitoring system can detect unusual activities while automatically obscuring personal details like faces and clothing. By balancing safety with privacy, TeD-SPAD demonstrates how smart city technologies can be designed to respect civil liberties.
M3Act
M3Act creates virtual scenarios of people walking, talking, and moving together in groups - think of it as a sophisticated computer simulation of crowd behavior. This technology helps develop better AI systems for understanding how people naturally interact in public spaces, with applications for urban safety and design.
Constellation
By collecting bird's-eye view images of urban intersections, the Constellation project provides vital data for creating AI systems that can help prevent accidents and improve traffic flow. This privacy-preserving approach, using high-mounted cameras that can't identify individuals, helps cities develop better safety systems for everyone sharing the road.
Thrust 3
Cookie Monster
Cookie Monster is a new approach to measuring online advertising that puts user privacy first while still allowing websites to understand their advertising effectiveness. It gives users clear visibility into how their data is being used and ensures their privacy is protected through mathematical safeguards built into their web browser.
DPack
DPack is an intelligent scheduling system that treats privacy as a limited resource, helping organizations analyze data while strongly protecting individual privacy. It maximizes the insights that can be gained from sensitive data while ensuring privacy guarantees are never compromised.
Thrust 5
Boundless
Boundless creates highly realistic virtual city environments to help develop AI systems that can better protect pedestrians and improve traffic flow at busy intersections. By using advanced computer simulation rather than real street cameras, this open-source platform helps cities enhance street safety while maintaining strong privacy protections for community members.