Abstract:
Techniques described herein relate to using reduced-dimensionality embeddings generated from robot sensor data to identify predetermined semantic labels that guide robot interaction with objects. In various implementations, obtaining, from one or more sensors of a robot, sensor data that includes data indicative of an object observed in an environment in which the robot operates. The sensor data may be processed utilizing a first trained machine learning model to generate a first embedded feature vector that maps the data indicative of the object to an embedding space. Nearest neighbor(s) of the first embedded feature vector may be identified in the embedding space. Semantic label(s) may be identified based on the nearest neighbor(s). A given grasp option may be selected from enumerated grasp options previously associated with the semantic label(s). The robot may be operated to interact with the object based on the pose and using the given grasp option.
Abstract:
Ein System und ein Verfahren zum Lernen bzw. Teachen eines Roboterbefehls, wobei eine Bedienperson (2) durch wiederholtes bzw. mehrfaches Handführen des Roboters (1), z.B. durch Drücken und/oder Ziehen mit variierenden Kräften an einem oder mehreren Robotergliedern, Roboterdaten erzeugt (S30), z.B. Positionen, Stellungen, Posen und/oder Kräfte, und wobei mittels eines neuronalen Netzes (6) durch Deep Learning auf Basis der zuvor erfassten Sequenz von Roboterdaten ein Beaufschlagungsmuster gelernt wird und einem Roboterbefehlt zugeordnet wird (S40).
Abstract:
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for a system configured to plan actions to be performed by a robotic agent interacting with an environment to accomplish an objective by determining an optimized trajectory of state - action pairs for accomplishing the objective. The system maintains a current optimized trajectory and a current trust region radius, and optimizes a localized objective within the current trust region radius of the current optimized trajectory to determine a candidate updated optimized trajectory. The system determines whether the candidate updated optimized trajectory improves over the current optimized trajectory. In response to determining that the candidate updated optimized trajectory improves over the current optimized trajectory, the system updates the current optimized trajectory to the candidate updated optimized trajectory and updates the current trust region radius.
Abstract:
Die vorliegende Erfindung betrifft ein Verfahren und eine Vorrichtung zur Festlegung eines Bewegungsablaufs für einen mehrachsigen Manipulator (M) eines Robotersystems, der mehrere, unterschiedliche Drehachsen bildende Glieder (G) und ein Endglied für ein Zusammenwirken mit einem Effektor (E) aufweist, wobei der Effektor (E) in einem Arbeitsraum (R) zumindest eine beliebige Operation durchführen soll, und wobei das Endglied des Manipulators (M) zur Durchführung der zumindest einen beliebigen Operation in eine beliebige Soll- Pose (x i ) in Bezug auf den Arbeitsraum (R) überführt werden soll, wobei der Manipulator (M) in mehreren Schritten (S i ;S j ) unter Annäherung des Endglieds an die Soll-Pose (x i ) bewegt und für jeden Schritt (S i ;S j ) zumindest ein definiertes Impedanzmuster (K x ) und/oder Admittanzmuster in Bezug auf zumindest eine Achse (A A ;A G ;A E ;A R ) festgelegt wird, die die Achse (A A ;A G ;A E ;A R ) eines mit dem Manipulator (M) verknüpften Koordinatensystems (C A ;C G ;C E ;C R ) bildet.
Abstract translation:
本发明涉及一种方法和用于确定导航中使用的运动序列的装置R具有用于导航用途为r的多个旋转成形构件(G)和一个端部构件的不同的轴的机器人系统的多轴操纵器(M) 具有相互作用,在一个工作空间中的执行器(e)中,其中所述效应(E)(R)的至少任何操作携带导航用途听到,且其中所述操纵器(M)用于执行导航的使用端构件引导所述至少一个任意的操作在一 应当传递相对于工作空间(R)的任何期望姿态(x i,i),其中操纵器(M)以几个步骤(S i, S j sub>)通过参考最终项和每个步骤(S )移动到目标姿态(x i sub> 小号<子>Ĵ子>)为至少一个所定义的阻抗图案(K <子> X 子>)和/或Admittanzmuster相对于至少一个轴线(一<子>一子>,一个<子 > 定义轴(A sub>; A sub>; 子>;一个<子>电子子>,一个 - [R 子>)pften坐标系(C <子>一子> C <子“G sub>; C E sub>; C R sub>)。 p>
Abstract:
Robotic manipulators may be used to manipulate objects. Manipulation data about manipulations performed on objects may be generated and accessed. This data may be analyzed to generate a profile indicating how an object may be manipulated. A portion of the profile may be transmitted to a particular robotic manipulator. For example, the portion may be based on a manipulation capability of the robotic manipulator. In turn, the robotic manipulator may use the portion of the profile to manipulate the object.
Abstract:
Data about a physical object in a real-world environment is automatically collected and labeled. A mechanical device is used to maneuver the object into different poses within a three-dimensional workspace in the real-world environment. While the object is in each different pose an image of the object is input from one or more sensors and data specifying the pose is input from the mechanical device. The image of the object input from each of the sensors for each different pose is labeled with the data specifying the pose and with information identifying the object. A database for the object that includes these labeled images can be generated. The labeled images can also be used to train a detector and classifier to detect and recognize the object when it is in an environment that is similar to the real-world environment.
Abstract:
This invention relates to force/torque sensor and more particularly to multi-axis force/torque sensor and the methods of use for directly teaching a task to a mechatronic manipulator. The force/torque sensor has a casing, an outer frame forming part of or connected to the casing, an inner frame forming part of or connected to the casing, a compliant member connecting the outer frame to the inner frame, and one or more measurement elements mounted in the casing for measuring compliance of the compliant member when a force or torque is applied between the outer frame and the inner frame.
Abstract:
A configurable robotic apparatus and system is remotely operable in difficult, hazardous, subterranean, or submerged environs. The apparatus merges diverse disciplines to effect inspecting, cleaning, treating, repairing or otherwise maintaining a wide variety of materials and conditions. Deployment environments include power, municipal water and wastewater plants, surface and submerged infrastructures (pipes, lines, conduits), and like industrial applications. Extensible and articulating modules, configurable through standardized and interchangeable connectors, provide unique flexibility, scalability and versatility to accommodate a wide range of shapes, surfaces, and obstacles. In-module intelligence and instrumentation eliminates the need for constant manual control through autonomous operation capable of simultaneous optimization and synchronization of multiple work processes, but manual override and remote control is provided to overcome unanticipated limitations. Benefits include improved efficiency, cost, and safety over prior art. High-performance, one-pass operation reduces facility downtime while incorporating environmentally responsible debris recovery.
Abstract:
The present invention relates to a robot using a learning control architecture comprising three layers: a reasoning layer, a reactive layer, and a modelling layer. The reasoning layer develops strategies from given commands and measured sensor/actuators signals, the reactive layer develop control commands from strategies and from sensor/actuator signals, and the modelling layer is used by both the reactive and reasoning layer to build a physical model of the world around the robot.
Abstract:
A method and apparatus for a robotic machining process that gives a controlled removal rate of material from a workpiece when an object, tool or workpiece, held by a robot is brought into contact with a stationary object, workpiece or tool. A signal indicative of the force applied by the held object t the stationary object is used to control the rate at which the robot moves the held object in relation to the stationary object. Associated with the robot is a controller that has tunable proportional and integral gains. The controller determines a command for the feed rate of the tool when the tool engages the workpiece. In response to that command, the proportional and integral gains are tuned to obtain a cutting force to be applied to the workpiece when the tool engages the workpiece that is substantially the same as a desired cutting force.