patent.fyi
Applications  /  US 2026/0356362
United States Patent Application Publication

US 2026/0356362 A1

Published
Nov 12, 2026
Filed
Feb 17, 2026
Earliest date on the face
May 7, 2025
Classification
B60W 30/165
First published here
Nov 20, 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.

1.3Fixed logic for following control

What ifWhat if fixed digital logic performed the following calculations and mode transitions without executing stored instructions?
Claim language“when executed by the one or more processors”

Dedicated arithmetic circuits calculate projected path coordinates and compare them with object coordinates at each sensor update. A hardwired state machine selects the following or fallback mode, and output registers hold the corresponding steering and acceleration requests.

1.4Path from camera lane geometry

What ifWhat if the controller generated the expected trajectory from camera-derived lane geometry without using yaw rate?
Claim language“based on a yaw rate”

A forward camera supplies lane boundary points that the controller projects onto a ground plane using camera calibration. The controller fits a curve to the midpoint between the boundaries and samples the curve at distances set by vehicle speed and a two-second prediction horizon.

1.5Deviation from target motion direction

What ifWhat if the controller detected front-vehicle divergence by comparing motion directions without using the front vehicle's position?
Claim language“based on position information of the front vehicle”

Imaging radar estimates the front vehicle's ground-relative velocity vector from Doppler measurements across multiple reflections and compensates for the host vehicle's motion. The controller compares the vector direction with the expected path tangent at a fixed look-ahead distance and flags divergence when the angular difference exceeds a threshold for 300 milliseconds.

1.6Target lane assignment from markings

What ifWhat if the controller used lane markings?
Claim language“position information of a surrounding vehicle adjacent to the front vehicle”

A camera estimates the host lane boundaries, and the controller compares the front vehicle's footprint with those boundaries. The controller drops the target once its footprint lies entirely beyond a boundary for three consecutive updates and otherwise retains it.

1.6Periodic following target selection

What ifWhat if the controller decided whether to retain the front vehicle on a fixed schedule without waiting for a trajectory deviation?
Claim language“when the front vehicle deviates from the expected traveling trajectory”

Every 50 milliseconds, the controller ranks forward vehicle tracks by lane occupancy and time headway. It switches to a higher-ranked track after that track leads the ranking for five consecutive updates and enters speed regulation when no eligible track remains.

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

Following control for warehouse tow tractors

An indoor tow tractor uses a yaw-rate sensor, wheel encoders, and a scanning lidar to follow another tractor through warehouse aisles. Its controller integrates measured yaw rate and wheel speed to project the tractor's path and tracks the lead tractor and adjacent traffic in the same coordinate frame. When the lead tractor leaves the projected path, the controller maintains following if the lead tractor returns toward the path and the predicted swept areas of the tractor and its trailers remain clear of adjacent traffic. Otherwise, the controller requests a controlled stop.

Component substitution

Yaw rate from differential wheel speeds

The controller uses rear wheel encoders and the measured rear track width as its yaw-rate source. It subtracts the left rear wheel's linear speed from the right rear wheel's linear speed and divides the difference by the track width. The controller filters this rate and integrates it with the mean rear wheel speed to project the ego vehicle's path. When the lead vehicle departs from that path, the controller uses radar tracks of the lead and adjacent vehicles to continue following while their predicted separation stays above a configured clearance and ends following when it falls below that clearance.

Parameter or scale variation

Speed dependent prediction and confirmation windows

A controller updates its path projection and vehicle tracks at 20 Hz. It integrates wheel speed and measured yaw rate over a 1.0-second horizon at speeds up to 5 m/s and increases the horizon linearly to 2.5 seconds at 20 m/s, holding it there at higher speeds. The controller registers a lead vehicle departure when the measured center of the lead vehicle lies more than 0.6 m laterally from the projected path for three consecutive updates. After confirmation, it maintains following when the extrapolated outlines of the lead and adjacent vehicles stay at least 0.5 m apart over the active horizon.

Combination with existing systems

Integration with lane centering control

A lane centering controller fits a lane centerline from forward camera images, while a following controller projects the ego vehicle's path from measured yaw rate and speed. When the lead vehicle departs from the projected path, the following controller projects the positions of the lead vehicle and adjacent vehicles onto the camera-derived centerline. It keeps the lead vehicle as the longitudinal target if the lead vehicle is returning toward the lane center and no adjacent vehicle is predicted to enter the gap between the ego and lead vehicles. The lane centering controller supplies steering commands, and the following controller supplies acceleration and braking requests.

Automated or data-driven variant

Learned classification of lead vehicle departures

A gradient-boosted decision tree classifier runs on a vehicle processor alongside the yaw-rate path estimator and the vehicle tracker. Its inputs include the lead vehicle's lateral offset from the projected path, the change in that offset, and the longitudinal and lateral gaps to the nearest adjacent vehicle. Training uses recorded track sequences labeled according to whether the lead vehicle returns to the projected path within two seconds or continues into another lane. After a detected departure, the controller keeps following when the estimated return probability exceeds 0.8 and the predicted gap to the adjacent vehicle exceeds a configured minimum.

Following decisions from track uncertainty

An extended Kalman filter fuses radar range and range rate with camera bearing to estimate the positions, velocities, and state covariance of the lead and adjacent vehicles. The controller detects lead vehicle departure by comparing the filtered position with the path projected from ego speed and yaw rate. After departure, it propagates both vehicle tracks for 1.5 seconds in 0.1-second increments and expands each vehicle footprint along the principal axes of its position covariance by two standard deviations. It maintains following when the expanded footprints remain disjoint at every prediction step and releases the lead vehicle target when they overlap.

Specific version of a part

Error-correcting flash instruction storage

Embedded flash stores the trajectory prediction and following-control instructions. An error-correcting memory controller corrects a single-bit error in a stored word during a read and signals a detected double-bit error.

Lockstep cores on a memory bus

Two processor cores fetch instructions from shared flash through a memory bus and execute the same control program in lockstep. A hardware comparator checks the cores' outputs and raises a fault when they differ.

Constant yaw rate path prediction

The processor samples forward speed and yaw rate, then integrates a constant-speed, constant-yaw-rate motion model in 100-millisecond steps over a two-second horizon. At near-zero yaw rate, it projects a straight path along the current vehicle heading.

Cross-track deviation with persistence

The processor projects the front vehicle's estimated center onto the nearest segment of the predicted path and computes the lateral distance to that segment. It reports a deviation when the distance exceeds 0.8 meters for three consecutive 50-millisecond updates.

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 10,059,336 B2
Published
Aug 28, 2018
Filed
Jan 6, 2017
Class
B60W 30/10
Document BUS 2025/0117010 A1
Published
Apr 10, 2025
Filed
Oct 4, 2023
Class
G05D 1/0214
Claim 1
Document A
Document B
1.1a memory configured to store at least one instruction; and
Memory 240, and/or a secondary storage device associated with controller 202, may store data and/or software instructions that may assist controller 202 in performing various functions, such as the functions of method 300 of FIG. 3.
B · US 2025/0117010 A1 · ¶ 22
1.2one or more processors connected to the memory,
The adaptive control system includes one or more processors and a memory communicably coupled to the one or more processors and storing a deviation module and a control module.
A · US 10,059,336 B2 · ¶ 5
Controller 202 may include memory 240 and one or more processors 245.
B · US 2025/0117010 A1 · ¶ 22
1.3wherein the at least one instruction is configured, when executed by the one or more processors, to cause the autonomous vehicle control device to:
The deviation module includes instructions that when executed by the one or more processors cause the one or more processors to generate expected inputs for controlling the vehicle along a segment of a roadway on which the vehicle is traveling by analyzing a present context of the vehicle using a driver model.
A · US 10,059,336 B2 · ¶ 5
Processor 245 may be configured to execute the software instructions.
B · US 2025/0117010 A1 · ¶ 22
1.4generate an expected traveling trajectory of the autonomous vehicle based on a yaw rate of the autonomous vehicle being controlled to follow a front vehicle;
In one embodiment, the deviation module 220 initially generates the preferred trajectory as part of, for example, generating the predicted route at 310. The preferred trajectory 260 is, for example, a queue of expected inputs for controlling the vehicle 100 along the segment of the roadway on which the vehicle 100 is presently traveling.
A · US 10,059,336 B2 · ¶ 40
Method 300 further includes, at step 304, determining an expected travel trajectory for autonomous mobile machine 12 to perform the task command.
B · US 2025/0117010 A1 · ¶ 26
1.5determine whether the front vehicle deviates from the expected traveling trajectory based on position information of the front vehicle;
Method 300 may further include, at step 310, identifying an incident limiting factor responsible for the deviation between the actual travel trajectory and the expected travel trajectory of mobile machine 12.
B · US 2025/0117010 A1 · ¶ 32
1.6determine, when the front vehicle deviates from the expected traveling trajectory, whether to maintain following control to follow the front vehicle based on the position information of the front vehicle and position information of a surrounding vehicle adjacent to the front vehicle, and
1.7control the autonomous vehicle based on a result of the determination of whether to maintain the following control.
The control module includes instructions that when executed by the one or more processors cause the one or more processors to control the vehicle based, at least in part, on the expected inputs when the deviation score satisfies a deviation threshold indicating that the received inputs are inadequate to maintain the vehicle along the preferred trajectory for the segment of the roadway.
A · US 10,059,336 B2 · ¶ 5

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

Statements in these documents
This is because the circumstances within which the assistive technologies operate do not prevent the driver from guiding the vehicle into circumstances of potential concern altogether.
A · US 10,059,336 B2 · ¶ 3
Thus, if the deviation score indicates that the driver is not providing adequate inputs for maintaining the preferred trajectory, then the adaptive control system can modify the received inputs to improve/maintain a path of the vehicle.
A · US 10,059,336 B2 · ¶ 4
Such productivity losses may be difficult to summarize and quantify as conventional techniques analyze the performance of the LHD at the end of each cycle through post-processing of data.
B · US 2025/0117010 A1 · ¶ 2
Therefore, it would be advantageous to identify factors effecting efficiently of the machine in substantially real time.
B · US 2025/0117010 A1 · ¶ 2
Cite aspatent.fyi, US 2026/0356362 A1, claim 1, first published 2026-11-20T00:27:46-05:00, https://patent.fyi/US20260356362/, SHA-256 649c6578ac91c138635c229501c808d4a9912cbf67dae642a11016d62d99bb41
First published
Permanent addresshttps://patent.fyi/US20260356362/
Content fingerprint (SHA-256)649c6578ac91c138635c229501c808d4a9912cbf67dae642a11016d62d99bb41
Methodv2026-11-18.2 · automated, not reviewed by a person
Correctionscorrections@patent.fyi