Our maintenance team is out very early this morning. We don’t want to attract too much attention when we send our ANTS up the skyscrapers.
The "Autonomous Networked Systems” that climb up the glass facades like geckos are energy self-sufficient: They recharge their batteries with the help of the wind that blows through their piezo-ceramic "hair.”
The mini-robots’ mission is to find thousands of surveillance cameras, replace their microchips with the latest versions, test the new software to ensure the cameras can make 3-D scans of people hurrying by far below them, and then compare the images with those stored in a database.
If there is a high degree of similarity with a person who is being searched for, that individual will be marked on screens at the security headquarters and the cameras will then automatically follow the person's path.
Security will play an important role in the networked world of 2050, and it won't only be limited to the danger of terrorism: Energy, transport, and water management systems will have to be protected against hacker attacks.
In the field of video surveillance, human beings are not able to analyze the flood of images from thousands of cameras. The best computers today can already identify abandoned suitcases at airports and cars being driven in the wrong direction in tunnels. They can also distinguish between dogs and human intruders in a dark backyard at night.
However, they’re not yet very good at identifying criminals in video images — there are simply too many possible variations due to viewing angles, lighting, and changes to appearance (glasses, beards, etc.).
Still, if computing power increases a thousand-fold as is expected, and if the machines are equipped with the right software, the computers of tomorrow will at least be able to calculate the probability that an individual on video is actually a person who is being searched for.