Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for controlling a robot in accordance with a real-time robotics control framework. One of the methods includes: receiving a definition of a custom real-time control function; and repeatedly executing the custom real-time control function at each predetermined tick of a real-time robotics system driving the robot, including: obtaining sensor measurements, computing a new position for an end effector of the robot based on the sensor measurements in order to satisfy a distance range specified by the custom real-time control function, computing new robot control signals to cause the robot to move the end effector to the new position, and providing the new robot control signals to the robot.

Abstract: Device and method for operating a robot. As a function of a first state of the robot and/or its surroundings and as a function of an output of a first model, a first part of a manipulated variable for activating the robot for a transition from the first state into a second state of the robot is determined. A second part of the manipulated variable is determined as a function of the first state and regardless of the first model. A quality measure is determined as a function of the first state and of the output of the first model using a second model. A parameter of the first model is determined as a function of the quality measure. A parameter of the second model is determined as a function of the quality measure and a setpoint value. The setpoint value is determined as a function of a reward.

Abstract: A method for planning a manipulation task of an agent, particularly a robot. The method includes: learning a number of manipulation skills wherein a symbolic abstraction of the respective manipulation skill is generated; determining a concatenated sequence of manipulation skills selected from the number of learned manipulation skills based on their symbolic abstraction so that a given goal specification indicating a given complex manipulation task is satisfied; and executing the sequence of manipulation skills.

Abstract: Device and method for operating a robot. As a function of a first state of the robot and/or its surroundings and as a function of an output of a first model, a first part of a manipulated variable for activating the robot for a transition from the first state into a second state of the robot is determined. A second part of the manipulated variable is determined as a function of the first state and regardless of the first model. A quality measure is determined as a function of the first state and of the output of the first model using a second model. A parameter of the first model is determined as a function of the quality measure. A parameter of the second model is determined as a function of the quality measure and a setpoint value. The setpoint value is determined as a function of a reward.

Abstract: A method for planning a manipulation task of an agent, particularly a robot. The method includes: learning a number of manipulation skills wherein a symbolic abstraction of the respective manipulation skill is generated; determining a concatenated sequence of manipulation skills selected from the number of learned manipulation skills based on their symbolic abstraction so that a given goal specification indicating a given complex manipulation task is satisfied; and executing the sequence of manipulation skills.