The Governors Highway Safety Association reports that 5984 pedestrians were killed in 2017 by cars or other moving vehicles, a 33-year high, while other kinds of traffic deaths have decreased. This is a growing problem in the U.S. as more and more pedestrians and drivers are distracted by their cellphones.
Researchers at Purdue University have devised a new way to alert distracted walkers — via the very smartphone that is distracting them — about oncoming vehicles, potentially reducing the number of pedestrian deaths.
The Purdue team has developed a system called PHADE — private human addressing — that allows surveillance and other cameras in public to send an alert directly to a smartphone.
The client side of PHADE is built on the Samsung Galaxy S4 smartphone, which logs accelerometer, gyroscope, and gravity readings at 100 Hz. On the server side, two Samsung Galaxy S5 smartphones serve as IP cameras to record and stream videos (15 fps, 2000 kb/s, 800 × 480 pixels) to the server. The server consists of two PCs with dual NVIDIA GTX 1080 Ti SLI.
“Currently a user may download our app and receive messages,” said Purdue researcher and Ph.D. student Siyuan Cao. “We design our system in a general way so that it can work as an additional layer in the existing communication protocol. But a smartphone manufacturer can also choose to embed it into the smartphones’ system. In that case, any pedestrian with a cellphone will receive the alerts.”
Courtesy of Purdue University.
Traditional data transmission protocols must first learn the destination’s IP or MAC address. The PHADE system instead uses motion patterns as the address code for communication. He Wang, an assistant professor in the Purdue Department of Computer Science, told Photonics Media that the smartphones then locally make their own decisions on whether to accept a message.
“Surveillance cameras are widely deployed. Human or AI systems can retrieve numerous types of information from the video feeds of these cameras,” said Wang. “Our technology enables public cameras to send customized messages to targets without any prior registration. Moreover, the address code used by the camera is under carefully designed transformation to protect the targets’ privacy.”
Cameras on the street can remind pedestrians about potentially dangerous situations such as an approaching car, theft, tailing, or an uncovered well.
“This system basically allows surveillance cameras to talk to the public through their individual phones,” Cao said. “The camera can send an almost instant alert to a pedestrian that a car is coming.”
The pedestrian would receive a message to their phone that reads: “Danger! Oncoming vehicle.”
The PHADE system works using a server to receive video streams from cameras to track people. The camera builds a packet by linking a message to the address code and broadcasts the packet.
Upon receiving the packet, a mobile device of each of the targets uses sensors to extract its owner’s behavior and follow the same transformation to derive a second address code. If the second address code matches with the address code in the message, the mobile device automatically delivers the message to its owner.
“Our technology serves as a bridge to connect cameras and people,” Wang said. “Surveillance cameras are widely deployed today and human and AI systems can retrieve numerous bits of information from the video feeds of these cameras. Our innovation turns that information into life-saving applications.”
Surveillance camera and security companies would be able to integrate the technology into their products directly as a key feature. The technology can also be used in shopping malls, museums, and other locations to send personalized messages to people without compromising their privacy.