Back to results
US 11,969,898 B2Granted
Multi-sensor optimization of automatic machines to prevent safety issues
AssigneeDatalogic IP Tech S.R.L.
InventorsFrancesco D'Ercoli, Marco Cumoli, Claudio Saporetti
Life of this patent
Estimating the term…
Abstract
A predictive system and process that predicts safety system activation in industrial environments when collaborative robots (COBOTs), automated guidance vehicles (AGVs), and other robots (individual or collectively “robots”) are interacting (i) between one another or (ii) between a robot and human. As provided herein, the predictive system is not meant to substitute traditional safety systems, but rather to detect and classify robot-to-robot and robot-to-human interactions and potential interactions thereof so as to limit or avoid those interactions altogether, thereby increasing safety and efficiency of the robots.
Claims19
A system for safely operating robots in an environment, said system comprising:
- a set of heterogeneous input sensors configured to generate sensed data related to humans and robots operating within the environment, the set of heterogeneous input sensors positioned off-board the robots;
- one or more safety systems being executed by the respective robots, and configured to slow or stop the robots in the event of the one or more safety systems detecting a potential interaction with another robot or human, the one or more safety systems including one or more input sensors each being different from the set of heterogeneous input sensors;
- and
- at least one processor in communication with the set of heterogeneous input sensors, and configured to:
- determine a probability that a safety system will activate as a function of the sensed data generated by at least one of the heterogeneous input sensors;
- preprocess, using a plurality using a plurality of data pre-processor sub-modules with at least two being different from one another, the sensed data by one or more of the input sensors to generate preprocessed sensed data;
- extract features, using a plurality of feature extraction sub-modules with at least two being different from one another, from the preprocessed sensed data so as to generate extracted feature data;
- classify, using a plurality of classifier sub-modules with at least two being different from one another, the extracted feature data to generate classified risk predictions indicative of a safety system activation;
- generate, using an ensemble classifier, an overall safety prediction that defines a probability of activation of a safety system in the area being monitored by the input sensors based on the classified risk predictions, the overall safety prediction being generated by executing a supervised learning algorithm;
- and
- generate an output signal to cause at least one of a notification to alert a human or a control command to alter planned movement of a robot to avoid activation of the safety system of the robot that causes the robot to slow down.
8 more elements
A method for safely operating robots in an environment, the method comprising:
- receiving sensed data from a set of heterogeneous sensors in the environment related to humans and robots operating within the environment, at least one of the heterogeneous input sensors positioned off-board the robots;
- slowing or stopping the robots, by one or more safety systems (i) being respectively executed by the robots and (ii) including input one or more sensors that are different from the set of heterogeneous input sensors, in the event of the robot detecting a potential interaction with another robot or human;
- determining a probability that a safety system will activate as a function of the received sensed data from at least one of the heterogeneous sensors;
- pre-processing, using a plurality of data pre-processor sub-modules with at least two being different from one another, the sensed data to generate preprocessed sensed data;
- extracting features, using a plurality of feature extraction sub-modules with at least two being different from one another, from the preprocessed sensed data so as to generate extracted feature data;
- classifying, using a plurality of classifier sub-modules with at least two being different from one another, extracted feature data to generate classified risk predictions indicative of a safety system activation;
- and
- generating, using an ensemble classifier, an overall safety prediction that defines a probability of activation of a safety system in the area being monitored by the input sensors based on the classified risk predictions, the overall safety prediction being generated by executing a supervised learning algorithm;
- and
- generating an output signal to cause at least one of a notification to alert a human or a control command to alter planned movement of a robot to avoid activation of the safety system of the robot that causes the robot to slow down.
7 more elements
Dependencies are inferred from the claim text.