The present invention proposes an inverse kinematics modeling and solving principle for multi-axis systems based on axis invariant, including: the D-H and D-H parameter determination principle based on fixed axis invarian, “Ju-Gibbs” quaternion and class direction cosine matrix principle, the inverse solution principle of general 6R and 7R robotic arms based on axial invariant. These principles are versatile, convenient, and precise. They can be set up as circuits, code, directly or indirectly, partially or completely within a multi-axis robot system. In addition, the present invention also includes analysis verification system constructed on these principles for designing and verifying multi-axis robot systems.

A robot manipulator including limbs moveable via bearings controlled by actuators; sensors to capture a bearing position and a bearing torque/bearing force; a first sensor to capture a force screw W; a housing downstream of the first sensor; a second sensor to capture a user force applied to the housing and/or a user torque; a computing unit to determine, using a dynamics model of the robot manipulator and based on particular bearing torque/bearing force, the force screw W, and the user force and/or the user torque, a first force and/or a first torque to shift the limbs and a second force and/or a second torque to apply to an external object via an effector, wherein the dynamics model includes at least gravitational forces and inertial forces based on the bearing position; and a storage unit to store the first and/or the second force, and/or the first and/or the second torque.

Methods, apparatus, and computer-readable media for determining and utilizing corrections to robot actions. Some implementations are directed to updating a local features model of a robot in response to determining a human correction of an action performed by the robot. The local features model is used to determine, based on an embedding generated over a corresponding neural network model, one or more features that are most similar to the generated embedding. Updating the local features model in response to a human correction can include updating a feature embedding, of the local features model, that corresponds to the human correction. Adjustment(s) to the features model can immediately improve robot performance without necessitating retraining of the corresponding neural network model.

The present disclosure relates to a method for generating a novel impedance configuration for a three-degree-of-freedom (3DOF) leg of a hydraulically-driven legged robot. The method includes: separately determining variations of input signals of an inner position-based control loop and an inner force-based control loop of a hydraulic drive unit of each joint based on an obtained mathematical model; generating a novel impedance configuration in which position-based control is performed on a hydraulic drive unit of a hip joint, and force-based control is performed on hydraulic drive units of a knee joint and an ankle joint in a hydraulic drive system of the leg of a to-be-controlled robot; and performing forward calculation by using the leg mathematical model, to obtain an actual position and a force variation of the foot of the leg of the to-be-controlled robot to control motion of the foot of the to-be-controlled robot within motion space.

Disclosed herein a method and apparatus for learning a locally-adaptive local device task based on cloud simulation. According to an embodiment of the present disclosure, there is provided a method for learning a locally-adaptive local device task. The method comprising: receiving observation data about a surrounding environment recognized by a local device; performing a domain randomization based on the observation data and a failure type of a task assigned to the local device and relearning a policy network of the assigned task based on the domain randomization; and updating a policy network of the local device for the assigned task by transmitting the relearned policy network to the local device.

A grasp generation technique for robotic pick-up of parts. A database of solid or surface models is provided for all objects and grippers which are to be evaluated. A gripper is selected and a random initialization is performed, where random objects and poses are selected from the object database. An iterative optimization computation is then performed, where many hundreds of grasps are computed for each part with surface contact between the part and the gripper, and sampling for grasp diversity and global optimization. Finally, a physical environment simulation is performed, where the grasps for each part are mapped to simulated piles of objects in a bin scenario. The grasp points and approach directions from the physical environment simulation are then used to train neural networks for grasp learning in real-world robotic operations, where the simulation results are correlated to camera depth image data to identify a high quality grasp.

An industrial robot motion accuracy compensation method includes: establishing a motion parameter database, wherein the motion parameter database includes a plurality of different reference operating conditions and a motion parameter of the industrial robot corresponding to each reference operating condition, and each reference operating condition is formed by combining each element in each set in a total set of operation conditions; acquiring a current operating condition of the industrial robot; determining whether there is a reference operating condition matched with the current operating condition in the motion parameter database; if yes, taking a motion parameter corresponding to the matched reference operating condition as a motion parameter corresponding to the current operating condition; if no, performing an interpolation on a motion parameter corresponding to the current operating condition, and taking an interpolation result as the motion parameter corresponding to the current operating condition.

The present disclosure provides a biologically-inspired robotic device comprising: a first member; a second member pivotably connected to the first member; one or more actuators; and a coupler/decoupler mechanism (CDC) selectively coupling or decoupling of the one or more actuators to the second member, such that, when the one or more actuators are coupled to the second member, the one or more actuators act to pivot the second member relative to the first member.

The present disclosure relates to utilizing idle processing resources of a robot to reduce future burden on such processing resources. In particular, idle processing resources are utilized to identify future scenarios, and generate reactions to such future scenarios. The generated reactions are stored, and quickly retrieved as needed if corresponding identified future scenarios occur.

A computer-implemented method for determining a control trajectory for a robotic device. The method includes: performing an information theoretic model predictive control applying a control trajectory sample prior in each time step to obtain a control trajectory for a given time horizon; determining the control trajectory sample prior depending on a data-driven trajectory prediction model which is trained to output a control trajectory sample as the control trajectory sample prior based on an actual state of the robotic device.