Enabling smart and safe communities through AI
Chelsea Cavlovic, Karen Lightman · 2019
When we see a potential collision about to happen, we will do just about anything to avoid it, and sometimes, through skill, quick thinking or blind luck, we do. We’ve all experienced those near misses – the close calls. Maybe we’re a pedestrian in a crosswalk who skips out of the way of a vehicle, or the driver who swerves to miss that car making a Pittsburgh left. These stories are often told throughout the day, but they are rarely reported. However, these stories can have much deeper value if the underlying cause could be understood and remedied, so that future collisions can be avoided and lives can be saved. CMU researchers are deploying a “City Scale Computing” AI-enabled pilot project in a real-world laboratory on the busy streets of Pittsburgh to help tell this story and reduce (and potentially eliminate) near misses. This project, supported in part by CMU’s Metro21: Smart Cities Institute, is capturing and processing high resolution video to facilitate computer vision algorithms research. In coordination with their partners, researchers are also using the test bed to address privacy, power management, data-management, and policy challenges associated with dense, high-resolution urban video capture. Their research will inform and enable future connected vehicle to infrastructure (V2X) deployments and address issues such as blind spots and other safety-related challenges. This session, presented in a moderated session format, brought together a dynamic panel of CMU faculty, along with their municipal and industry partners to explore how scientific data can empower people and affect policy. Panelists will discuss how public-private partnerships can help solve real world problems, as well as inform the future states of artificial intelligence and connected vehicles.