Abstract: A waste sorting robot gripper comprises a suction cup engageable with the surface of a waste object. The suction cup has an air hole for evacuating air from the suction cup. A suction tube is coupled to the suction cup. The suction tube comprises a longitudinal axis. A first air inlet is in fluid communication with the air hole at one end of the suction tube and an air outlet at the other end of the suction tube. A path of the air flow between the air inlet and the air outlet is substantially along the longitudinal axis. The suction tube comprises a second air inlet in fluid communication with an air source, the second air inlet being between the first air inlet and the air outlet.

Abstract: Feedback information is an important aspect in all machine learning systems. In robot systems that are picking objects from a plurality of objects this has been arranged by acquiring images of objects that have been picked. When images are acquired after picking they can be imaged accurately and the information about picking and also dropping success can be improved by using acquired images as a feedback in the machine learning arrangement being used for controlling the picking and dropping of objects.

Abstract: A waste sorting robot can include a manipulator comprising a suction gripper for interacting with one or more waste objects to be sorted within a working area, and wherein the manipulator is moveable within the working area. There is a controller configured to send control instructions to the manipulator. At least one pressure sensor is in fluid communication with the suction gripper and configured to generate a pressure signal in dependence on a fluid pressure in the suction gripper. The controller is configured to receive the pressure signal and to determine manipulator instructions in dependence on the pressure signal.