Needless to say, cameras on self-driving cars cannot see around the corners of buildings.However, the ORCa computer vision system could one day allow them to do so… with a little help from shiny objects, they able look.
ORCa, whose name is an acronym for “Objects as Radiance-Field Cameras,” was developed by scientists at MIT and Rice University.
In a nutshell, the technology analyzes the distorted reflections of shiny surfaces for sharper images What is being reflected, and how far away it is. In a self-driving car scenario, the shiny surface of another car parked near an upcoming intersection (to name just one example) could provide a view of the approaching vehicle at the intersection.
The system works by taking multiple images of the shiny object in question, each taken from a slightly different angle as the camera moves closer to the object. Machine-learning-based software breaks down reflections from an object’s surface into individual pixels.
By analyzing how those pixels change relative to one image, the software is able to determine the shape of an object (along with its color and texture), compensating for the way it distorts reflections.
Furthermore, by modeling the reflective scene as a so-called 5D radiation field, ORCa is able to determine the direction and intensity of the light rays that strike or emit at each point exist That information in turn allows the system to determine the distance of each point from the reflective surface and from each other.
“We’ve shown that any surface can be turned into a sensor by this formula of turning objects into virtual pixels and virtual sensors,” said MIT graduate student Kushagra Tiwary, co-lead author of a study. Paper academically. “This can be applied in many different fields.”
In addition to applications such as collision avoidance systems for self-driving cars, the technology could also be used in drones. By analyzing reflections from smooth objects on the ground, a drone can obtain a ground-based view of the environment in which it is flying.
source: and
