# US 2026/0097486 A1 · Techniques for training robots for dexterous manipulation tasks > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 2026/0097486 A1 - Document ID: US20260097486A1 - Status: Published application - Kind: A1 - Application number: 19/348,334 - Filed: 2025-10-02 - Published: 2026-04-09 - Earliest claimed: 2024-10-04 - Assignees: Massachusetts Institute of Technology - Applicants: Massachusetts Institute of Technology - Inventors: Haoshu Fang; Arthur Hu; Branden Romero; Edward Adelson; Pulkit Agrawal - Web page: https://patentgrep.com/patent/US20260097486A1 ## CPC - B25J 9/0081: {with leader teach-in means} - B25J 9/0006: {Exoskeletons, i.e. resembling a human figure} - B25J 9/1612: {characterised by the hand, wrist, grip control} - B25J 13/082: {Grasping-force detectors (in general G01L5/16, G01L5/22)} - B25J 13/088: {with position, velocity or acceleration sensors} - B25J 19/021: {Optical sensing devices} - B25J: MANIPULATORS; CHAMBERS PROVIDED WITH MANIPULATION DEVICES ({manipulators specially adapted for use in surgery A61B34/70; manipulators used in cleaning hollow articles B08B9/04}; manipulators associated with rolling mills B21B39/20; manipulators associated with forging machines B21J13/10; {manipulators associated with picking-up and placing mechanisms B23P19/007}; means for holding wheels or parts thereof B60B30/00; {vehicles with ground-engaging propulsion means, e.g. walking members B62D57/02, B62D57/032; devices for picking-up and depositing articles or materials between conveyors B65G47/90, B65G47/91; manipulators with gripping or holding means for transferring packages B65H67/065}; cranes B66C; {manipulators used in the protection or supervision of pipe-line installations F17D5/00; walking equipment adapted for nuclear steam-generators F22B37/006}; manipulators specially adapted for, or associated with, nuclear reactors G21C; {apparatus used for handling wafers during manufacture or treatment of semiconductor H10P72/50}) ## Abstract The techniques described herein relate to systems, apparatus, articles of manufacture, and methods for training robots for dexterous manipulation tasks. An example apparatus for training a robotic system includes an end effector configured to manipulate an object, a passive exoskeleton configured to be manipulated by a portion of a user's body to change one or more parameters of the end effector, one or more linkages coupling the end effector to the passive exoskeleton, wherein the one or more linkages transmit forces applied to the end effector, due to interactions of the end effector with a surrounding environment comprising the object, to the portion of the user's body through the passive exoskeleton, and one or more sensors coupled to the end effector and configured to sense the one or more parameters of the end effector, and output one or more signals indicative of the one or more parameters. ## Claims ### Claim 1 An apparatus for training a robotic system comprising: an end effector configured to manipulate an object; a passive exoskeleton configured to be manipulated by a portion of a user's body to change one or more parameters of the end effector; one or more linkages coupling the end effector to the passive exoskeleton, wherein the one or more linkages transmit forces applied to the end effector, due to interactions of the end effector with a surrounding environment comprising the object, to the portion of the user's body through the passive exoskeleton; and one or more sensors coupled to the end effector and configured to: sense the one or more parameters of the end effector; and output one or more signals indicative of the one or more parameters. ### Claim 2 (depends on 1) The apparatus of claim 1, further comprising one or more hardware processors configured to: receive the one or more signals indicative of the one or more parameters; and generate a training dataset for operation of the robotic system using the one or more parameters of the end effector. ### Claim 3 (depends on 2) The apparatus of claim 2, wherein the end effector comprises one or more actuators operatively coupled to the end effector and configured to change the one or more parameters of the end effector in accordance with the training dataset. ### Claim 4 (depends on 1) The apparatus of claim 1, wherein the end effector comprises a plurality of fingers configured to manipulate the object. ### Claim 5 (depends on 1) The apparatus of claim 1, wherein the one or more linkages are one or more rigid linkages. ### Claim 6 (depends on 1) The apparatus of claim 1, wherein the one or more parameters of the end effector comprise at least one of an acceleration, a direction, a pose, or a speed of a portion of the end effector, and the one or more sensors are configured to sense the at least one of the acceleration, the direction, the pose, or the speed of the portion of the end effector. ### Claim 7 (depends on 6) The apparatus of claim 6, wherein the one or more sensors coupled to the end effector comprise one or more encoders and/or one or more inertial measurement units coupled to the end effector, and the one or more encoders and/or the one or more inertial measurement units is/are configured to: sense the at least one of the acceleration, the direction, the pose, or the speed of the portion of the end effector; and output the one or more signals indicative of the at least one of the acceleration, the direction, the pose, or the speed of the portion of the end effector. ### Claim 8 (depends on 1) The apparatus of claim 1, further comprising: a limb connected to the end effector; and one or more second sensors coupled to the limb and configured to: sense one or more second parameters of the limb; and output one or more second signals indicative of the one or more second parameters. ### Claim 9 (depends on 1) The apparatus of claim 1, wherein the one or more parameters comprise a force applied by a portion of the end effector to the object, the one or more sensors coupled to the end effector comprise one or more tactile sensors disposed on one or more contact pads of the end effector, and the one or more tactile sensors are configured to: sense the force applied by the portion of the end effector to the object; and output the one or more signals indicative of at least the sensed force. ### Claim 10 (depends on 1) The apparatus of claim 1, further comprising one or more image sensors configured to: capture a stream of images of at least one of the end effector, the object, or the surrounding environment during manipulation of the object; and output the stream of images. ### Claim 11 (depends on 1) The apparatus of claim 1, wherein the passive exoskeleton does not comprise an actuator. ### Claim 12 A method for training a robotic system comprising: manually manipulating, using a passive exoskeleton, one or more parameters of an end effector to manipulate an object; transmitting, using one or more linkages of the passive exoskeleton, forces applied to the end effector to a portion of a user's body due to interactions of the end effector with a surrounding environment; and sensing, using one or more sensors, one or more parameters of the end effector while facilitating manipulation of the object to generate a training dataset for operation of the robotic system. ### Claim 13 (depends on 12) The method of claim 12, further comprising: receiving, using at least one communication interface, one or more signals indicative of the one or more parameters of the end effector; and generating, using one or more hardware processors and the one or more parameters, the training dataset for operation of the robotic system. ### Claim 14 (depends on 13) The method of claim 13, further comprising: generating, using the one or more hardware processors and the training dataset, machine readable instructions for operation of the robotic system in accordance with the training dataset; and outputting, using the at least one communication interface, the machine readable instructions to the robotic system for operation of the robotic system. ### Claim 15 (depends on 12) The method of claim 12, further comprising operating the robotic system in accordance with the training dataset and using the one or more parameters during manipulation of the object or a different object. ### Claim 16 (depends on 12) The method of claim 12, further comprising: capturing, using at least one image sensor, a stream of images of at least one of the end effector, the object, or the surrounding environment during manipulation of the object; and generating, using one or more hardware processors and the stream of images, the training dataset for operation of the robotic system. ### Claim 17 At least one computer-readable storage medium storing processor executable instructions that, when executed by at least one hardware processor, cause the at least one hardware processor to: receive, using at least one communication interface, sensor data from one or more sensors configured to sense one or more parameters of an end effector when the end effector is manually manipulated by a passive exoskeleton to manipulate an object; generate, using the sensor data, a training dataset for operation of a robotic system comprising the end effector; and cause a change in the one or more parameters of the end effector to manipulate the object in accordance with the training dataset. ### Claim 18 (depends on 17) The at least one computer-readable storage medium of claim 17, wherein the processor executable instructions cause the at least one hardware processor to: receive one or more signals indicative of the one or more parameters of the end effector; and generate, using the one or more parameters, a training dataset for operation of the robotic system. ### Claim 19 (depends on 18) The at least one computer-readable storage medium of claim 18, wherein the processor executable instructions are first processor executable instructions and cause the at least one hardware processor to: generate second processor executable instructions for operation of the robotic system in accordance with the training dataset; and output the second processor executable instructions to the robotic system for operation of the robotic system. ### Claim 20 (depends on 17) The at least one computer-readable storage medium of claim 17, wherein the one or more parameters comprise at least one of motion data, pose data, or force data associated with the end effector, and the processor executable instructions cause the at least one hardware processor to: receive, from at least one image sensor, a stream of images of at least one of the end effector, the object, or an environment during manipulation of the object; and generate, using the at least one of the motion data, the pose data, the force data, or the stream of images, the training dataset for operation of the robotic system.