Poisson-guided auto scanning sounds complicated, but what it really represents is the actualization of an idea that a robot could scan an object more accurately than a human, which is not a novel or amazing notion anymore. 3D Scanning physical data, such as objects with unusual or complicated geometry, is no easy task, and there has been an industry-wide effort to reduce the amount of damaged data and blind spots that make 3D images incorrect or incomplete. One team has come up with results that are perfectly in line with the notion of benevolent automation.
From the Visual Computing Research Center, Tel-Aviv University, the Memorial University of Newfoundland, the University of Konstanz and Shandong University, a team of researchers has created a series of experiments using an Artec Spider scanner in the hand and arm of an anthropomorphic robot named PR2. In their tests, the PR2, armed with the Artec Spider, scanned a number of small objects. The robot held and rotated a small resin table that carried and spun each object in place.
The scanning process works like this: First an initial scan is performed, in order to capture an initial point cloud that roughly covers the object’s surface. The Poisson equation generates a set of candidate viewpoints, which informs the robot on possible next scans. But first the robot moves the scanner to record snapshots from these viewpoints. The robot’s hand holding the scanner moves into position based on its assigned viewpoint and a scan is made. The system obtains the frame, which is then registered and merged with the initial image.
This whole scanning process was programmed with Artec’s Scanning SDK. A basic way to interpret this is that the robot is capable of taking a lot more scans, more quickly and more accurately than you or anyone else. The scanning also takes place automatically and stops once a specified reconstruction requirement is reached.
The group of researchers set out on an experiment to discover a way to create high fidelity automatic scanning of objects, and they succeeded. See for yourself:
