Methods, systems, and apparatus, including computer programs encoded on computer storage media, for enhanced path planning. In some implementations, a first path is determined for travel by a robot, the first path extending from an origin to a destination. Path segments are determined based on the first path. A corner between two of the path segments has an angle less than a predetermined threshold is determined. In response to determining that the corner between two of the path segments has an angle less than the predetermined threshold, a bypass path segment is determined that bypasses the corner. A second path for the robot to travel is determined based on the path segments and the bypass path segment. Data indicating the second path is provided to the robot.

A method for controlling a robot device. The method includes performing an initial training of an actor neural network by imitation learning of demonstrations, controlling the robot device by the initially trained actor neural network to generate multiple trajectories, wherein each trajectory comprises a sequence of actions selected by the initially actor neural network in a sequence of states, and observing the return for each of the selected actions, performing an initial training of a critic neural network by supervised learning, wherein the critic neural network is trained to determine the observed returns of the actions selected by the initially actor neural network, training the actor neural network and the critic neural network by reinforcement learning starting from the initially trained actor neural network and the initially trained critic neural network and controlling the robot device by the trained actor neural network and trained critic neural network.

Methods, systems, and apparatus, including computer programs encoded on computer storage media, for generating a trajectory of a robot. One of the methods includes receiving a plurality of path points; processing each network input in an input sequence that is derived from the path points using a trajectory generation neural network to generate an output sequence comprising a plurality of network outputs, each network output specifying a respective displacement between two adjacent trajectory points; and generating, based on the output sequence, a predicted trajectory of the robot.

Parcel deploying apparatus including package receiving and/or delivering apparatus defining an enclosure and any or all walls of the enclosure being releasably locked to the package receiving or deploying apparatus by an externally controllable locking mechanism and movable from a closed to an open orientation defining a package receiving flat surface. A robot including a body with controllably movable apparatus for aligning movements, the body including a wall releasably attached to the body for movement between a body closing orientation and a horizontal orientation defining a package receiving flat surface. One of the enclosure and the body including package relocating apparatus which moves a package between a stored and an extended orientation on one of the flat surfaces. The controllably movable apparatus controllable to move the robot into alignment with a delivery position. The package receiving and/or delivering apparatus and the robot can be integrated into a single robosafe.

A method for controlling a robot device. The method includes performing an initial training of an actor neural network by imitation learning of demonstrations, controlling the robot device by the initially trained actor neural network to generate multiple trajectories, wherein each trajectory comprises a sequence of actions selected by the initially actor neural network in a sequence of states, and observing the return for each of the selected actions, performing an initial training of a critic neural network by supervised learning, wherein the critic neural network is trained to determine the observed returns of the actions selected by the initially actor neural network, training the actor neural network and the critic neural network by reinforcement learning starting from the initially trained actor neural network and the initially trained critic neural network and controlling the robot device by the trained actor neural network and trained critic neural network.

A time-optimal trajectory generation method, for a robotic manipulator having a transport path with at least one path segment, comprising generating a forward time-optimal trajectory of the manipulator along the at least one path segment from a start point of the at least one path segment towards an end point of the at least one path segment, generating a reverse time-optimal trajectory of the manipulator along the at least one path segment from the end point towards the start point of the at least one path segment, and combining the time-optimal forward and reverse trajectories to obtain a complete time-optimal trajectory, where the forward and reverse trajectories of the at least one path segment are blended together with a smoothing bridge joining the time-optimal forward and reverse trajectories in a position-velocity reference frame with substantially no discontinuity between the time-optimal forward and reverse trajectories.

The present disclosure provides a motion control method and apparatus and a robot with the same. The method includes: obtaining a first rotational angle P.sub.1 of an output shaft of the servo currently at and a first time T.sub.1 for the output shaft of the servo to perform one rotation; obtaining a second rotational angle P.sub.2 for the output shaft of the servo to reach and a second time T.sub.2 for the output shaft of the servo to rotate from the first rotational angle P.sub.1 to the second rotational angle P.sub.2; calculating a motion curve B(t) of the output shaft of the servo based on the first rotational angle P.sub.1, the second rotational angle P.sub.2, the first time T.sub.1, and the second time T.sub.2; and controlling the servo to rotate according to the motion curve B(t). The present disclosure solves the instability in the gravity center of the robot.

Parcel deploying apparatus including package receiving and/or delivering apparatus defining an enclosure and any or all walls of the enclosure being releasably locked to the package receiving or deploying apparatus by an externally controllable locking mechanism and movable from a closed to an open orientation defining a package receiving flat surface. A robot including a body with controllably movable apparatus for aligning movements, the body including a wall releasably attached to the body for movement between a body closing orientation and a horizontal orientation defining a package receiving flat surface. One of the enclosure and the body including package relocating apparatus which moves a package between a stored and an extended orientation on one of the flat surfaces. The controllably movable apparatus controllable to move the robot into alignment with a delivery position. The package receiving and/or delivering apparatus and the robot can be integrated into a single robosafe.

A numerical controller includes a reading analysis unit that reads a CNC program and additional information, a path generation unit that determines a movement path of a tool, and a velocity control unit that determines a velocity for moving the tool according to the movement path of the tool, and machining errors, deterioration of a machined surface quality, or an increase in a cycle time are reduced without increasing a CNC program size and a calculation time associated with control more than necessary.

The present disclosure provides a motion control method and apparatus and a robot with the same. The method includes: obtaining a first rotational angle P.sub.1 of an output shaft of the servo currently at and a first time T.sub.1 for the output shaft of the servo to perform one rotation; obtaining a second rotational angle P.sub.2 for the output shaft of the servo to reach and a second time T.sub.2 for the output shaft of the servo to rotate from the first rotational angle P.sub.1 to the second rotational angle P.sub.2; calculating a motion curve B(t) of the output shaft of the servo based on the first rotational angle P.sub.1, the second rotational angle P.sub.2, the first time T.sub.1, and the second time T.sub.2; and controlling the servo to rotate according to the motion curve B(t). The present disclosure solves the instability in the gravity center of the robot.