patent.fyi
Applications  /  US 2026/0355959
United States Patent Application Publication

US 2026/0355959 A1

Published
Nov 12, 2026
Filed
May 6, 2025
Earliest date on the face
May 6, 2025
Classification
B25J 9/1671
First published here
Nov 19, 2026

Generated automatically. Not reviewed by a person. Not legal advice or an opinion.

Section 1 · machine-generated

What if?

Questions about how individual parts of claim 1 might be changed. A question may describe something different from what the claim language recites, or a version of it; no view is offered on which.

1.1Planning from explicit action rules

What ifWhat if a rule-based planner calculated the motion from the target geometry at run time?
Claim language“accessing a latent codebook comprising a vector”

An action table stores a rotation axis, a requested turn angle and a grip force limit for the turn operation. The planner uses the target surface model to select opposing grip points and calculates successive wrist poses about the target axis.

1.2Geometry from tactile probing

What ifWhat if a touch probe supplied the geometry of the second object?
Claim language“an image of a second object”

A robot moves a spring-loaded contact stylus along programmed approach directions and records the robot pose at each trigger. The controller converts the poses into surface coordinates and fits the dimensions and axis needed for the gripping operation.

1.4Contact planning on target geometry

What ifWhat if the planner calculated the contact trajectory directly on the second object's surface?
Claim language“modifying a first dimensionality associated with the first object”

The image supplies a surface model of the second object, and the stored vector supplies the action identity. The planner selects fingertip contacts on the target surface, advances the desired object pose in angular increments and solves inverse kinematics at each increment to maintain contact.

Section 2 · machine-generated

Variations

Further arrangements of the same subject matter, described here as of the publication date at the foot of this page.

Other field of use

Contact transfer for greenhouse plug trays

A greenhouse transplanter stores a latent vector that encodes fingertip contact trajectories for lifting a seedling plug from a reference tray cell. A calibrated overhead image locates the corners of a target tray cell, and the tray recipe supplies the cell depth and taper. A decoder scales the lateral offsets to the cell opening and adjusts the insertion offsets to follow the taper. The controller outputs finger approach, closure, and lift waypoints for the target cell.

Component substitution

Surface mapping with thin plate splines

A thin plate spline replaces uniform dimensional scaling when transferring a contact trajectory between molded housings. The decoder reconstructs the source contact samples and their times from the selected latent vector, and a calibrated image supplies target landmarks such as rim corners and fastener recesses. A regularized spline fitted to paired source and target landmarks maps the contact samples onto the target geometry. The controller projects each mapped sample onto the reconstructed target surface and outputs the resulting waypoints in time order.

Parameter or scale variation

Contact transfer for miniature assembly parts

A miniature assembly station handles rigid parts 2 to 8 millimeters wide with a parallel microgripper. A calibrated overhead camera samples the fixture plane at 10 micrometers per pixel, and a fixture datum supplies the reference height. The decoder scales the stored contact offsets independently along the two in-plane axes using the measured part width and length. The controller interpolates the mapped contacts into setpoints at 5 millisecond intervals and limits approach speed to 2 millimeters per second.

Combination with existing systems

Contact transfer on an indexed conveyor

An indexed pallet conveyor carries fixtures through an imaging station and a manipulation station. A pallet reader retrieves the action identifier, and a calibrated camera measures the target object dimensions and pose relative to the pallet. The trajectory generator decodes the selected latent vector, maps the contact samples into pallet coordinates, and stores the resulting waypoints under the pallet identifier. At the manipulation station, a locating stop seats the pallet and the controller transforms the stored waypoints into manipulator coordinates.

Automated or data-driven variant

Image based fitting of deformable fixtures

A trained keypoint detector locates the rim, slots, and corner tabs of a flexible fixture in synchronized images from two calibrated cameras. Software triangulates the keypoints and fits a deformable surface mesh whose vertices include the corresponding reference landmarks. Each contact decoded from the selected latent vector is stored as a triangle identifier and barycentric weights on the reference mesh. The trajectory generator applies those weights to the fitted mesh and outputs contact waypoints with approach directions derived from local surface normals.

Tactile correction of transferred contact paths

An image supplies the target dimensions for the initial mapping of a decoded contact trajectory, and a tactile sensor array on each finger measures the contact location during execution. The controller converts each pressure centroid into object coordinates using the measured gripper pose and compares it with the scheduled contact point. It applies the resulting tangential position correction to the next three contact setpoints and adjusts finger closure with a separate normal-force loop. The corrected positions and force targets form the instruction stream for the remaining motion.

Specific version of a part

Action encoded as object motion

The codebook vector stores an object-relative rotation axis and a series of target angles. A grasp planner selects opposing surface patches on the target and calculates wrist poses that carry the object through those angles.

Action selected from fixture state

A clamp switch reports that the fixture holds the object, and a displacement sensor measures the height of a partly seated cap. The controller selects its turn operation when the clamp is closed and the cap height exceeds the seating threshold.

Reactive control from terminal targets

The controller receives a target rotation angle and a fingertip force setpoint. At each control cycle, it uses the measured object angle and fingertip forces to compute wrist velocity and jaw effort, then stops wrist motion when the angle error falls within the specified tolerance.

Latent vector for paired contacts

An encoder compresses 32 time samples of thumb and index fingertip contact coordinates into a 16-component vector stored in the codebook. A decoder reconstructs the ordered contact coordinates in a frame fixed to the first object.

Calibrated color and depth image

A fixed camera captures registered color and depth channels for the second object. The controller segments the object and uses the camera calibration to express its visible surface coordinates in the robot base frame.

Axis scaling for cylindrical targets

The controller estimates the target cylinder's axis, radius and height from the depth image. It decodes the stored contact samples, multiplies their radial and axial coordinates by the respective target-to-source radius and height ratios, then places the scaled samples in the target cylinder's coordinate frame.

Section 3 · quoted from USPTO records

Other work

Passages reproduced word for word from two earlier-published U.S. patent documents, placed beside the language of claim 1. No characterization is offered. Each passage is marked A or B for the document it comes from; ¶ numbers are the paragraph numbers in the USPTO’s published full text of that document.

Document AUS 2023/0278218 A1
Published
Sep 7, 2023
Filed
Feb 22, 2023
Class
B25J 9/1676
Document BUS 2025/0121502 A1
Published
Apr 17, 2025
Filed
Aug 18, 2024
Class
B25J 9/1669
Claim 1
Document A
Document B
1.1accessing a latent codebook comprising a vector, wherein the vector describes a specific action that is performed by a first gripping structure against a first object, wherein the vector describes the specific action by mapping a trajectory of contact points between the first gripping structure and the first object during performance of the specific action;
1.2accessing an image of a second object that is to be manipulated, wherein the second object is different from the first object;
The camera 12 is configured to acquire images of the target site 16, which may be presented in different forms, including but not limited to, captured still images, real-time images and/or digital video signals. “Real-time” as used herein generally refers to the updating of information at the same rate as data is received.
A · US 2023/0278218 A1 · ¶ 29
The screen of the device may display the image received from the gripper camera, which may be angled to show a target object.
B · US 2025/0121502 A1 · ¶ 25
1.3obtaining an instruction indicating that the specific action is to be performed against the second object;
The user may interact with the screen via selecting the target object as shown in the screen, indicating to the manipulator to initiate a grasp of the target object.
B · US 2025/0121502 A1 · ¶ 25
1.4in response to accessing the vector in the latent codebook, using the image to localize the vector to the second object, wherein localizing the vector to the second object includes modifying a first dimensionality associated with the first object, as described in the vector, to account for a second dimensionality of the second object, as identified by the image, such that a new trajectory of new contact points between a second gripping structure and the second object are generated; and
In accordance with the preferred embodiment of the present invention, grasp candidates are synthesized using the image of the object and the segmented 3D map of the target object.
B · US 2025/0121502 A1 · ¶ 29
1.5generating a sequence of instructions that represent the specific action as applied to the second object, wherein the sequence of instructions identifies the new trajectory of the new contact points between the second gripping structure and the second object, such that the sequence of instructions facilitates performance of the specific action against the second object.
Lastly, the controller C is programmed to transmit the command via one or more signals to the appropriate joint motor of the robotic arm 24 and/or the coupling plate 26 according to the movement sequence. The transmitted commands cause motors at the respective joints to move the robotic arm 24 and/or the coupling plate 26, thereby causing the camera 12 to move while avoiding collisions.
A · US 2023/0278218 A1 · ¶ 44

An empty cell means only that no passage was reproduced there.

Statements in these documents
As such, it may be challenging to avoid self-collisions of the multiple components, for example, one robot link with another, collisions of the camera unit with the robotic arm and/or storage unit, and collisions of the camera unit with the image plane.
A · US 2023/0278218 A1 · ¶ 2
In some embodiments, the predefined buffer zone is within a delta value of at least one keep-out zone, application of the trajectory scaling factor pushing the camera away from the at least one keep-out zone.
A · US 2023/0278218 A1 · ¶ 4
Fully autonomous grasp synthesis still faces challenges in terms of adaptability and reliability, especially in dynamic and unstructured environments.
B · US 2025/0121502 A1 · ¶ 3
In such systems, the cognitive load to the user and time required for execution is minimized through partial automation, while the user is presented with visualizations and the option to guide or intervene throughout the process.
B · US 2025/0121502 A1 · ¶ 3
Cite aspatent.fyi, US 2026/0355959 A1, claim 1, first published 2026-11-19T17:33:15-05:00, https://patent.fyi/US20260355959/, SHA-256 65208b83278d58d5a1e441b6fe45c6d197308abf84496796466b08d4836cb3a6
First published
Permanent addresshttps://patent.fyi/US20260355959/
Content fingerprint (SHA-256)65208b83278d58d5a1e441b6fe45c6d197308abf84496796466b08d4836cb3a6
Methodv2026-11-18.2 · automated, not reviewed by a person
Correctionscorrections@patent.fyi