Faster and more accurate way has been developed by roboticists so that robots are able to grasp objects quickly and precisely, including in changing and fluttered environments period this new development has the potential to enhance the usefulness of robots not only in domestic settings but also in industrial sector. Through the implementation of this new approach, robots are able to scan an environment quickly and map each pixel grasp quality using a depth image. High accuracy rates have been achieved in real world tests.
Doctor Jürgen Leitner said that basic human tasks such as picking and grasping an object is extremely difficult for machines so being able to successfully program the robots in very controlled environments to grasp and pick up very specific items is an achievement. The existing robots shortcomings including the ability to changes when an object is moved.
As environment is constantly changing and get mixed up and these changes often happen without any warning, it is essential for robots to be able to adapt to these changes and work in unstructured environments so as to be useful. The new method developed makes use of an approach the works by forecasting the quality and pose of 2 fingered grasp at every pixel.