An automated robotic assembly system is configured to assemble a device by fitting a first component made of a material liable to deformation by external force with a second component by means of a robot, and the first component is provided with a distortion detection unit for detecting distortion thereof. If the distortion detected by the distortion detection unit exceeds a predetermined value, a signal for notifying abnormality is output to stop an automated assembly operation by the robot.

Embodiments herein generally relate to methods, systems and devices for automated assembly of building structures. In at least one embodiment, there is provided a method for automated assembly of building structures, the method comprises analyzing assembly data associated with a building structure; based on the analyzing, determining an assembly sequence for assembling building parts to construct the building structure, wherein the assembly sequence comprises a plurality of assembly tasks; generating robot-specific control instructions, for each of one or more assembly robots in a robotic assembly cell, to execute the assembly sequence; and transmitting the robot-specific control instructions to the one or more assembly robots in the robotic assembly cell.

Techniques are disclosed for training and applying machine learning models to control robotic assembly. In some embodiments, force and torque measurements are input into a machine learning model that includes a memory layer that introduces recurrency. The machine learning model is trained, via reinforcement learning in a robot-agnostic environment, to generate actions for achieving an assembly task given the force and torque measurements. During training, experiences are collected as transitions within episodes, the transitions are grouped into sequences, and the last two sequences of each episode have a variable overlap. The collected transitions are stored in a prioritized sequence replay buffer, from which a learner samples sequences to learn from based on transition and sequence priorities. Once trained, the machine learning model can be deployed to control various types of robots to perform the assembly task based on force and torque measurements acquired by sensors of those robots.

A computer-implemented method of determining actions for controlling a robot, in particular an assembly robot, includes (i) receiving a first and second input, wherein the first input is a sentence describing an action which should be carried out by the robot, wherein the second input is an image of a current state of an environment of the robot, (ii) feeding the first input into a first machine learning model and feeding the second input into a second machine learning model, wherein the first and second machine learning models are configured to determine tokens for their respective inputs, and (iv) feeding the tokens into a third machine learning model, wherein the third machine learning model outputs two outputs, wherein the first output is a switch for incorporating specialized skill networks and the second output are actions.

A method for controlling a robotic manipulator according to a task comprises accepting a feedback signal including a sequence of multi-modal observations of a state of execution of the task. The multi-modal observations are processed with a neural network having a self-attention module with a hierarchically conditioned output to produce a skill of the robotic manipulator and an action conditioned on the skill. The neural network is trained in a supervised manner with demonstration data to produce a sequence of skills and a corresponding sequence of actions for the actuators of the robotic manipulator to perform the task. The method further comprises determining one or more control commands for the one or more actuators based on the produced action and submitting the one or more control commands to the one or more actuators causing a change of the state of execution of the task.

A method for controlling a robot to perform a task. The method includes acquiring, for each target of a sequence of targets comprising at least one intermediate target of the task and a final target of a task, a target image data element comprising at least one target image from a perspective of an end-effector of the robot at a respective target position of the robot and successively according to the sequence of targets, for each target in the sequence, acquiring, for the target, an origin image data element, supplying the origin image data element and the target image data to a machine learning model configured to derive a delta movement between the origin current position and the target position and controlling the robot to move according to the delta movement.

A manufacturing matrix can include a plurality of work cells. The manufacturing matrix can include a plurality of robotic arms disposed in the plurality of work cells to produce a construction product. The manufacturing matrix can include a storage location to store inventory including at least one of the material, the tool, the subassembly of the construction product, or a completed construction product. The manufacturing matrix can include a transportation system to move the inventory within the manufacturing matrix. The manufacturing matrix can include a data processing system communicably coupled with the plurality of robotic arms and the transportation system. The data processing system can provide a first instruction to a first robotic arm cell and a second instruction to a second robotic arm of the second work cell.

Embodiments herein generally relate to methods, systems and devices for automated assembly of building structures. In at least one embodiment, there is provided a method for automated assembly of building structures, the method comprises analyzing assembly data associated with a building structure; based on the analyzing, determining an assembly sequence for assembling building parts to construct the building structure, wherein the assembly sequence comprises a plurality of assembly tasks; generating robot-specific control instructions, for each of one or more assembly robots in a robotic assembly cell, to execute the assembly sequence; and transmitting the robot-specific control instructions to the one or more assembly robots in the robotic assembly cell.

Embodiments of the present disclosure provide a pre-welding addressing method and system. The pre-welding addressing method includes: obtaining product information of a to-be-addressed battery pack, in response to the battery pack getting in position, where the battery pack includes a plurality of cells, and the product information includes initial addressing coordinates of a plurality of cell terminal posts; obtaining positioning deviation values of the plurality of cell terminal posts based respectively on the initial addressing coordinates of the plurality of cell terminal posts, where the positioning deviation values of the cell terminal posts are determined based on addressing images of the cell terminal posts that are acquired by an addressing camera module; and determining target addressing coordinates of the plurality of cell terminal posts based respectively on the initial addressing coordinates of the plurality of cell terminal posts and the positioning deviation values of the plurality of cell terminal posts.

Provided is an automotive glass setting apparatus and a method for setting a vehicle glass. The method includes loading a glass having a plurality of edges onto a setting base, scanning, by a scanning unit, the plurality of edges of the loaded glass, and calculating a center of the glass based on data scanned by the scanning unit. The method further includes adjusting, by a plurality of alignment units and a plurality of moving mechanisms, a position of the glass to align the calculated center of the glass with a center of the setting base.