patent.fyi
Applications  /  US 2026/0355988
United States Patent Application Publication

US 2026/0355988 A1

Published
Nov 12, 2026
Filed
Jun 21, 2026
Earliest date on the face
Dec 22, 2023
Classification
B25J 19/007
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.

1.1Records keyed by joint

What ifWhat if the storage indexed measured reducer characteristics by robot joint position and contained no reducer type identifiers?
Claim language“information enabling a type of a speed reducer to be specified”

A commissioning routine measures lost motion and torsional stiffness at each assembled joint and writes the results to a fixed array ordered from base to wrist. The simulator reads coefficients from the array position corresponding to each joint.

1.1Direct measurement during analysis

What ifWhat if the analysis used live measurements from a reducer test stand with no stored characteristic records?
Claim language“a storage that stores”

A drive motor applies a torque sweep to the reducer while input and output encoders measure angular displacement and a torque transducer measures output load. The solver consumes the synchronized signals as they arrive to update the simulated joint response.

1.2Fixed robot dynamics model

What ifWhat if the device used a permanently wired robot model with no robot specification input?
Claim language“receives specification information of a robot”

Fixed gains in an analog state model represent the effective inertias and stiffnesses of a two-joint test arm near a reference pose. Integrator circuits calculate joint angle and speed from the applied motor torque and the reducer response.

1.3Automatic reducer selection

What ifWhat if the device selected the reducers for all joints automatically with no reducer designation input?
Claim language“receives designation of a speed reducer”

The device calculates each joint’s peak torque and maximum speed over the specified motion cycle. A selector filters a component table by those limits and assigns the lowest-mass remaining reducer to each joint.

1.4Live healthy reference reducer

What ifWhat if a healthy reference reducer supplied the normal response directly while storage held only abnormal-state characteristics?
Claim language“stores the characteristic information of the speed reducer in a normal state”

A dynamometer applies the requested speed and load to a healthy reference reducer. The normal-case solver uses the measured output torque and angular displacement as they arrive, and the abnormal-case solver reads the corresponding characteristics from storage.

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

Reducer selection for greenhouse picking arms

A greenhouse picking station receives arm link dimensions, payload masses, and programmed moves from the picking controller. A setup screen assigns a reducer type to each joint, and a database stores stiffness and lost motion for healthy and worn conditions under each type identifier. A simulator runs the programmed approach and withdrawal moves with both sets of characteristics and records gripper position relative to the picking point.

Component substitution

Piecewise linear torque and deflection tables

Piecewise linear lookup tables replace coefficient-based reducer models in the state generator. The tables store measured output deflection at torque breakpoints for loading and unloading, with separate table sets indexed by reducer type and condition. At each simulation step, the generator selects the table for the assigned reducer and torque direction, interpolates the deflection, and applies it to the joint angle used in the link position calculation.

Parameter or scale variation

Single joint fault sweep for dual arms

A model contains two six-axis arms and receives a reducer type assignment for each of the 12 joints. The simulator runs one case with all reducers in their normal condition and 12 additional cases, each using abnormal characteristics at one joint and normal characteristics at the other 11 joints. Each run follows the same 20-second coordinated transfer and records both tool center positions at 2-millisecond intervals.

Combination with existing systems

Reducer models in an offline programming system

An offline robot programming system supplies the link geometry, payload properties, and time-stamped joint trajectory to a reducer state simulator. A joint configuration file maps each robot joint to a reducer type in a library of normal and abnormal characteristics. The simulator returns a tool pose trace for each condition, and the programming system displays both traces against the same fixture geometry.

Controller testing with a simulated robot

A physical robot controller connects to a real-time robot simulator over a cyclic industrial fieldbus. The simulator loads the robot geometry and joint reducer assignments, then resolves each assignment against a library of normal and abnormal reducer characteristics. It receives commanded joint torques and returns simulated motor and output shaft positions through the controller's drive feedback interface. The test rack executes the same motion program with each characteristic set and records the resulting tracking errors and controller stop signals.

Automated or data-driven variant

Sensor fitted reducer characteristic tables

A test fixture drives an identified reducer through repeated torque reversals while a torque transducer and shaft encoders record output torque and angular displacement. A fitting routine subtracts the input angle divided by the reduction ratio from the output angle and tabulates the resulting angular difference against torque for each reversal direction. The routine stores separate tables from unworn and worn specimens under the same reducer type identifier. The robot simulator retrieves both tables for each assigned joint and calculates the resulting link motion under the programmed load.

Specific version of a part

Reducer lookup by type

A flash memory table uses a reducer type code as the lookup key. Each record stores the reduction ratio and a torque–deflection curve represented by pairs of torque and angle values.

Robot specification file reader

A file reader imports link masses, inertia tensors, joint axes, and fixed transforms from a structured text file. It checks joint references and loads the values into the robot dynamics model.

Reducer menu per joint

A configuration panel presents one reducer selection menu for each joint. The receiver records each selected reducer type code with its joint identifier.

Paired torque reversal curves

Each reducer record contains two torque–deflection curves measured over the same load range. One curve represents a reducer with nominal clearances, and the other represents a unit with increased lost motion after wear.

Cite aspatent.fyi, US 2026/0355988 A1, claim 1, first published 2026-11-19T18:11:03-05:00, https://patent.fyi/US20260355988/, SHA-256 6ac06bee9ba3d6b66a24aa104223a4748ab95dea14dde082a846c880bd3d9099
First published
Permanent addresshttps://patent.fyi/US20260355988/
Content fingerprint (SHA-256)6ac06bee9ba3d6b66a24aa104223a4748ab95dea14dde082a846c880bd3d9099
Methodv2026-11-18.2 · automated, not reviewed by a person
Correctionscorrections@patent.fyi