# US 12,181,304 B2 · Method and device for building road model

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

- Number: US 12,181,304 B2
- Document ID: US12181304B2
- Status: Granted
- Kind: B2
- Application number: 17/582,162
- Filed: 2022-01-24
- Published: 2022-07-28
- Granted: 2024-12-31
- Earliest claimed: 2021-01-25
- Assignees: Nio Technology (Anhui) Co., Ltd
- Applicants: Nio Technology (Anhui) Co., Ltd
- Inventors: Siwei Peng
- Web page: https://patentgrep.com/patent/US12181304B2

## CPC

- G01C 21/3815: {Road data}
- G01C 21/30: Map- or contour-matching
- G01C 21/3804: {Creation or updating of map data}
- G01C 21/3841: {Data obtained from two or more sources, e.g. probe vehicles}
- G06V 10/74: Image or video pattern matching; Proximity measures in feature spaces
- G06V 20/588: {Recognition of the road, e.g. of lane markings; Recognition of the vehicle driving pattern in relation to the road}
- G01C: MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY (measuring liquid level G01F; radio navigation, determining distance or velocity by use of propagation effects, e.g. Doppler effects, propagation time, of radio waves, analogous arrangements using other waves G01S)
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING

## Abstract

The invention relates to a method for building a road model. The method includes: building a lane line pool, the lane line pool including one or more historical lane lines; updating the lane line pool based on matching between a current lane line and the one or more historical lane lines in the lane line pool; and performing lane matching on all lane lines in the updated lane line pool, so as to build the road model. The invention further relates to a device for building a road model, a computer storage medium, and a vehicle.

## Claims

### Claim 1

A method for building a road model, comprising:
building a lane line pool, the lane line pool comprising one or more historical lane lines;
generating, by a vision sensor, sensor data;
identifying a current lane line based on the sensor data;
updating the lane line pool based on matching between the current lane line and the one or more historical lane lines in the lane line pool;
performing lane matching on all lane lines in the updated lane line pool, so as to build the road model and avoid a real world misjudgment when the sensor misses detection of lane lines; and
calculating one or more autonomous driving functions for a vehicle based on the road model, the one or more autonomous driving functions being executed by the vehicle,
wherein performing lane matching on all lane lines in the updated lane line pool, so as to build the road model comprises:
building a lane pool, the lane pool comprising one or more historical lanes generated based on the one or more historical lane lines;
when a first lane line in the updated lane line pool matches a first historical lane of the one or more historical lanes, associating the first lane line with the first historical lane by updating lane line information corresponding to the first historical lane; and
when a second lane line in the updated lane line pool matches none of the one or more historical lanes, generating a new lane based on the second lane line, and re-establishing a topological relationship for left and right lanes of the new lane.

### Claim 2 (depends on 1)

The method of claim 1, wherein updating the lane line pool based on matching between the current lane line and the one or more historical lane lines in the lane line pool comprises:
calculating a matching score of the current lane line and a historical lane line of the one or more historical lane lines; and
when the matching score is higher than a first threshold, determining that the current lane line matches the historical lane line.

### Claim 3 (depends on 2)

The method of claim 2, wherein calculating the matching score of the current lane line and the historical lane line comprises:
calculating the matching score of the current lane line and the historical lane line based on an average distance between the current lane line and the historical lane line.

### Claim 4 (depends on 1)

The method of claim 1, wherein updating the lane line pool based on matching between the current lane line and the one or more historical lane lines in the lane line pool comprises:
calculating a matching score of the current lane line and a historical lane line of a plurality of historical lane lines included in the one or more historical lane lines; and
matching the historical lane line and the current lane line when the historical lane line has a highest matching score from among the plurality of historical lane lines.

### Claim 5 (depends on 1)

The method of claim 1, wherein updating the lane line pool based on matching between the current lane line and the one or more historical lane lines in the lane line pool comprises:
when the current lane line matches one of the one or more historical lane lines, updating lane line information corresponding to the current lane line in the lane line pool.

### Claim 6 (depends on 5)

The method of claim 5, wherein updating the lane line pool based on matching between the current lane line and the one or more historical lane lines in the lane line pool further comprises:
when the current lane line matches none of the one or more historical lane lines, adding the current lane line to the lane line pool.

### Claim 7 (depends on 5)

The method of claim 5, wherein updating the lane line pool based on matching between the current lane line and the one or more historical lane lines in the lane line pool further comprises:
estimating the one or more historical lane lines based on motion information; and
for a historical lane line in the lane line pool for which matching has not been completed, retaining an estimated state of the historical lane line and lowering a corresponding confidence level of the historical lane line.

### Claim 8 (depends on 7)

The method of claim 7, wherein when the confidence level is higher than a threshold, the historical lane line for which matching has not been completed is retained in the lane line pool.

### Claim 9 (depends on 1)

The method of claim 1, further comprising:
after performing lane matching on all lane lines in the updated lane line pool, so as to build the road model, deleting an invalid lane in the lane pool that does not match any lane line in the updated lane line pool.

### Claim 10 (depends on 9)

The method of claim 9, further comprising:
after performing lane matching on all lane lines in the updated lane line pool, so as to build the road model, for a valid lane in the lane pool, checking left and right lane line information of the valid lane; and
when one of the left and right lane line information is missing, calculating an attribute of a missing lane line associated with the missing one of the left and right lane line information based on a lane line associated with the other of the left and right lane line information and lane line information of an adjacent lane.

### Claim 11

A device for building and using a road model, comprising:
a memory; and
a computer processor configured to:
build a lane line pool, the lane line pool comprising one or more historical lane lines;
receive sensor data generated based on output of a vision sensor;
identify a current lane line based on the sensor data;
update the lane line pool based on matching between the current lane line and the one or more historical lane lines in the lane line pool;
perform lane matching on all lane lines in the updated lane line pool, so as to build the road model and avoid a real world misjudgment when the sensor misses detection of lane lines; and
calculate one or more autonomous driving functions for a vehicle based on the road model, the one or more autonomous driving functions being executed by the vehicle,
wherein performing lane matching on all lane lines in the updated lane line pool, so as to build the road model comprises:
building a lane pool, the lane pool comprising one or more historical lanes generated based on the one or more historical lane lines; and
generating a new lane by using a lane line in the updated lane line pool, or associating the lane line in the updated lane line pool with the one or more historical lanes;
wherein after performing lane matching on all lane lines in the updated lane line pool, so as to build the road model, the computer processor is further configured to:
delete an invalid lane in the lane pool that does not match any lane line in the updated lane line pool;
for a valid lane in the lane pool, check left and right lane line information of the valid lane; and
when one of the left and right lane line information is missing, calculate an attribute of a missing lane line associated with the missing one of the left and right lane line information based on a lane line associated with the other of the left and right lane line information and lane line information of an adjacent lane.

### Claim 12

A non-transitory computer storage medium, comprising instructions, wherein when the instructions are run, a method for building and using a road model is performed, the method comprising:
building a lane line pool, the lane line pool comprising one or more historical lane lines;
receiving sensor data generated based on output of a vision sensor;
identifying a current lane line based on the sensor data;
updating the lane line pool based on matching between the current lane line and the one or more historical lane lines in the lane line pool;
performing lane matching on all lane lines in the updated lane line pool, so as to build the road model and avoid a real world misjudgment when the sensor misses detection of lane lines; and
calculating one or more autonomous driving functions for a vehicle based on the road model, the one or more autonomous driving functions being executed by the vehicle,
wherein performing lane matching on all lane lines in the updated lane line pool, so as to build the road model comprises:
building a lane pool, the lane pool comprising one or more historical lanes generated based on the one or more historical lane lines;
when a first lane line in the updated lane line pool matches a first historical lane of the one or more historical lanes, associating the first lane line with the first historical lane by updating lane line information corresponding to the first historical lane; and
when a second lane line in the updated lane line pool matches none of the one or more historical lanes, generating a new lane based on the second lane line, and re-establishing a topological relationship for left and right lanes of the new lane.