# US 12,223,410 B2 · Lane selection using machine learning

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 12,223,410 B2
- Document ID: US12223410B2
- Status: Granted
- Kind: B2
- Application number: 18/589,391
- Filed: 2024-02-27
- Published: 2024-06-20
- Granted: 2025-02-11
- Earliest claimed: 2019-12-31
- Assignees: Google Inc.
- Applicants: Google LLC
- Inventors: Thomas Deselaers; Victor Carbune
- Web page: https://patentgrep.com/patent/US12223410B2

## CPC

- G06N 3/006: based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
- B60W 30/18163: {Lane change; Overtaking manoeuvres}
- G06F 18/217: {Validation; Performance evaluation; Active pattern learning techniques}
- G06N 3/08: Learning methods
- G06N 3/092: Reinforcement learning
- G06N 7/01: Probabilistic graphical models, e.g. probabilistic networks
- G06N 20/00: Machine learning
- G06V 10/764: using classification, e.g. of video objects
- G06V 10/82: using neural networks
- G06V 20/58: Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
- G06V 20/588: {Recognition of the road, e.g. of lane markings; Recognition of the vehicle driving pattern in relation to the road}
- B60W 2420/403: Image sensing, e.g. optical camera
- B60W 2520/10: Longitudinal speed
- B60W 2552/10: Number of lanes
- B60W 2555/20: Ambient conditions, e.g. wind or rain
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- B60W: CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING

## Abstract

To select a lane in a multi-lane road segment for a vehicle travelling on the road segment, a system identifies, in multiple lanes and in a region ahead of the vehicle, another vehicle defining a target; the system applies an optical flow technique to track the target during a period of time, to generate an estimate of how fast traffic moves; and the system applies the estimate to machine learning (ML) model to generate a recommendation which one of the plurality of lanes the vehicle is to choose.

## Claims

### Claim 1

A method for selecting a lane in a multi-lane road segment for a vehicle travelling on the road segment, the method comprising:
identifying, in a plurality of lanes and in a region ahead of the vehicle, another vehicle defining a target;
applying an optical flow technique to track the target during a period of time, to generate an estimate of how fast traffic moves; and
applying the estimate to machine learning (ML) model to generate a recommendation which one of the plurality of lanes the vehicle is to choose.

### Claim 2 (depends on 1)

The method of claim 1, further comprising:
receiving, from a camera mounted in the vehicle, imagery of a road segment during a certain period of time; and
applying image classification to the imagery to identify the target.

### Claim 3 (depends on 1)

The method of claim 1, further comprising:
receiving imagery of the region ahead of the vehicle from an external camera.

### Claim 4 (depends on 1)

The method of claim 1, further comprising:
processing a feedback signal subsequent to a lane change in accordance the recommendation, the feedback signal corresponding to one of a time or a distance travelled; and
applying the feedback signal to the ML model.

### Claim 5 (depends on 1)

The method of claim 1, further comprising:
identifying a plurality of respective targets in the plurality of lanes; and
applying the optical flow technique to track the targets, to generate respective estimates of how fast the plurality of targets move.

### Claim 6 (depends on 5)

The method of claim 5, further comprising:
applying, to the ML model, the estimates of how fast the plurality of targets move in the respective lanes as a reward for executing a lane change.

### Claim 7 (depends on 6)

The method of claim 6, wherein applying the estimates as the reward includes, for a lane change from an original lane to a new lane and subsequently to executing the lane change:
calculating a first reward based on the speed of travel in the original lane; and
calculating a second reward based on the speed of travel in the new lane.

### Claim 8 (depends on 1)

The method of claim 1, wherein the ML model includes a metric of discomfort associated with a lane change, the metric specific to a driver of the vehicle.

### Claim 9 (depends on 8)

The method of claim 8, further comprising:
generating the metric of discomfort for the driver based on driving patterns of the driver.

### Claim 10 (depends on 1)

The method of claim 1, further comprising:
determining a cost metric associated with a lane change in view of environmental conditions; and
using the cost metric with the ML model to generate the recommendation.

### Claim 11

A computing system configured to select a lane in a multi-lane road segment for a vehicle travelling on the road segment, the system comprising processing hardware and configured to:
identify, in a plurality of lanes and in a region ahead of the vehicle, another vehicle defining a target;
apply an optical flow technique to track the target during a period of time, to generate an estimate of how fast traffic moves; and
apply the estimate to machine learning (ML) model to generate a recommendation which one of the plurality of lanes the vehicle is to choose.

### Claim 12 (depends on 11)

The system of claim 11, further configured to:
receive, from a camera mounted in the vehicle, imagery of a road segment during a certain period of time; and
apply image classification to the imagery to identify the target.

### Claim 13 (depends on 11)

The system of claim 11, further configured to:
receive imagery of the region ahead of the vehicle from an external camera.

### Claim 14 (depends on 11)

The system of claim 11, further configured to:
process a feedback signal subsequent to a lane change in accordance the recommendation, the feedback signal corresponding to one of a time or a distance travelled; and
apply the feedback signal to the ML model.

### Claim 15 (depends on 11)

The system of claim 11, further configured to:
identify a plurality of respective targets in the plurality of lanes; and
apply the optical flow technique to track the targets, to generate respective estimates of how fast the plurality of targets move.

### Claim 16 (depends on 15)

The system of claim 15, further comprising:
apply, to the ML model, the estimates of how fast the plurality of targets move in the respective lanes as a reward for executing a lane change.

### Claim 17 (depends on 16)

The system of claim 16, wherein to apply the estimates as the reward the system is configured, for a lane change from an original lane to a new lane and subsequently to executing the lane change, to:
calculate a first reward based on the speed of travel in the original lane; and
calculate a second reward based on the speed of travel in the new lane.

### Claim 18 (depends on 11)

The system of claim 11, wherein the ML model includes a metric of discomfort associated with a lane change, the metric specific to a driver of the vehicle.

### Claim 19 (depends on 18)

The system of claim 18, further configured to:
generate the metric of discomfort for the driver based on driving patterns of the driver.

### Claim 20 (depends on 11)

The system of claim 11, further configured to:
determine a cost metric associated with a lane change in view of environmental conditions; and
use the cost metric with the ML model to generate the recommendation.