Handling assembly comprising a handling device for carrying out at least one work step, method, and computer program
AssigneeRobert Bosch GmbH
InventorsChristian Knoll, Corinna Pfeiffer-Scheer, Dieter Kunz, Jens Hofele, Peter Schlaich
Life of this patent
Estimating the term…
Abstract
A handling assembly having a handling device for carrying out at least one working step with and/or on a workpiece in a working region of the handling device, stations being situated in the working region, with at least one monitoring sensor for the optical monitoring of the working region and for provision as monitoring data, with a localization module, the localization module being designed to recognize the stations and to determine a station position for each of the stations.
Claims15
A handling assembly that operates using training data, the handling assembly comprising:
- a robotic arm;
- at least one monitoring sensor configured to optically monitor a working region and to provide monitoring data based on the optical monitoring;
- and
- a processor, wherein:
- the processor is configured to:
- based on the monitoring data, recognize stations that are situated in the working region and to determine a respective station position for each of the stations;
- interpret a user definition of a work step that is input as a semantic sentence user-formulated by a user-input of phrases that are (a) user-selected from a plurality of phrases that are predefined for the processor independently of one another, and (b) user-linked together in the semantic sentence with a specification of one or more of the stations;
- and
- based on the interpretation of the user definition of the work step, control the robotic arm to carry out the work step with and/or on a workpiece in the work region at the one or more of the stations;
- and
- the handling assembly has at least one of the following two features (a) and (b):
- (a) the processor is configured to execute a training using the training data to determine, based on the training data, recognition features as station recognition data for the recognition of the stations;
- and
- (b) the recognition of the stations is based on a machine learning of station features extracted from the training data, and the interpretation is based on the recognition of the stations.
A method for operating a robotic arm, the method using training data and the method comprising the following steps:
- optically monitoring, using a monitoring sensor, a working region to thereby generate monitoring data;
- providing, by the monitoring sensor, the monitoring data to a processor;
- based on the monitoring data, the processor recognizing stations that are situated in the working region and determining a respective station position for each of the recognized stations;
- interpreting, by the processor, a user definition of a work step that is input as a semantic sentence user-formulated by a user-input of phrases that are (a) user-selected from a plurality of phrases that are predefined for the processor independently of one another, and (b) user-linked together in the semantic sentence with a specification of one or more of the stations;
- and
- based on the interpretation of the user definition of the work step, controlling, by the processor, the robotic arm to carry out the work step with and/or on a workpiece in the work region at the one or more of the stations;
- wherein the method has at least one of the following two features (a) and (b):
- (a) the method further comprises executing, by the processor, a training using the training data to determine, based on the training data, recognition features as station recognition data for the recognition of the stations;
- and
- (b) the recognition of the stations is based on a machine learning of station features extracted from the training data, and the interpretation is based on the recognition of the stations.
A non-transitory computer-readable storage medium on which is stored a computer program having program code that is executable by a computer and that, when executed by the computer, causes the computer to perform a method that uses training data and that includes the following steps:
- optically monitoring, using a monitoring sensor, a working region;
- based on the monitoring, recognizing stations that are situated in the working region and determining a respective station position for each of the recognized stations;
- interpreting a user definition of a work step that is input as a semantic sentence user-formulated by a user-input of phrases that are (a) user-selected from a plurality of phrases that are predefined for a processor independently of one another, and (b) user-linked together in the semantic sentence with a specification of one or more of the stations;
- and
- based on the interpretation of the user definition of the work step, controlling a robotic arm to carry out the work step with and/or on a workpiece in the work region at the one or more of the stations;
- wherein the method has at least one of the following two features (a) and (b):
- (a) the method further comprises executing, by the processor, a training using the training data to determine, based on the training data, recognition features as station recognition data for the recognition of the stations;
- and
- (b) the recognition of the stations is based on a machine learning of station features extracted from the training data, and the interpretation is based on the recognition of the stations.
Dependencies are inferred from the claim text.