# US 2021/0239490 A1 · Generating Positions of Map Items for Placement on a Virtual Map

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

- Number: US 2021/0239490 A1
- Document ID: US20210239490A1
- Status: Published application
- Kind: A1
- Application number: 16/886,634
- Filed: 2020-05-28
- Published: 2021-08-05
- Earliest claimed: 2020-02-05
- Assignees: Electronic Arts Inc.
- Applicants: Electronic Arts Inc.
- Inventors: Han Liu; Yiwei Zhao; Jingwen Liang; Mohsen Sardari; Harold Chaput; Navid Aghdaie; Kazi Zaman
- Web page: https://patentgrep.com/patent/US20210239490A1

## CPC

- G01C 21/3673: {Labelling using text of road map data items, e.g. road names, POI names}
- G01C 21/3638: {including 3D objects and buildings (three dimensional [3D] modelling, e.g. data description of 3D objects G06T17/00; geographic models G06T17/05)}
- G01C 21/367: {Details, e.g. road map scale, orientation, zooming, illumination, level of detail, scrolling of road map or positioning of current position marker}
- G01C 21/3682: {output of POI information on a road map (G01C21/3614, G01C21/3685 take precedence)}
- G06F 16/29: Geographical information databases
- G06N 3/045: Combinations of networks
- G06N 3/0464: Convolutional networks [CNN, ConvNet]
- G06N 3/0475: Generative networks
- G06N 3/08: Learning methods
- G06N 3/084: Backpropagation, e.g. using gradient descent
- G06N 3/09: Supervised learning
- G06N 3/094: Adversarial learning
- 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)
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

This specification describes a system for generating positions of map items such as buildings, for placement on a virtual map. The system comprises: at least one processor; and a non-transitory computer-readable medium including executable instructions that when executed by the at least one processor cause the at least one processor to perform at least the following operations: receiving an input at a generator neural network trained for generating map item positions; generating, with the generator neural network, a probability of placing a map item for each subregion of a plurality of subregions of the region of the virtual map; and generating position data of map items for placement on the virtual map using the probability for each subregion. The input to the generator neural network comprises: map data comprising one or more channels of position information for at least a region of the virtual map, said one or more channels including at least one channel comprising road position information for the region; and a latent vector encoding a selection of a placement configuration.

## Claims

### Claim 1

A system for generating positions of map items for placement on a virtual map, comprising:
at least one processor; and
a non-transitory computer-readable medium including executable instructions that when executed by the at least one processor cause the at least one processor to perform at least the following operations:
receiving an input at a generator neural network trained for generating map item positions, the input comprising:
map data comprising one or more channels of position information for at least a region of the virtual map, said one or more channels including at least one channel comprising road position information for the region; and
a latent vector encoding a selection of a placement configuration;
generating, with the generator neural network, a probability of placing a map item for each subregion of a plurality of subregions of the region of the virtual map; and
generating position data of map items for placement on the virtual map using the probability for each subregion.

### Claim 2 (depends on 1)

The system of claim 1, wherein the map items comprise structures for placement on the virtual map.

### Claim 3 (depends on 2)

The system of claim 2, wherein the generator neural network is further configured to generate height data for at least some of said structures, comprising generating a height for one or more of said subregions.

### Claim 4 (depends on 1)

The system of claim 1, wherein the one or more channels further comprise at least one of:
a channel of waterway position information;
a channel of vegetation position information; and
a channel of existing map item position information.

### Claim 5 (depends on 1)

The system of claim 1, wherein the at least one channel of road position information comprises at least one of:
a channel of highway road position information;
a channel of arterial road position information; and
a channel of local road position information.

### Claim 6 (depends on 1)

The system of claim 1, wherein the latent vector is determined from an instance of map data.

### Claim 7 (depends on 1)

The system of claim 1, wherein the latent vector is sampled from a learned latent space.

### Claim 8 (depends on 1)

The system of claim 1, wherein the generator neural network is further configured to generate type information for the map items.

### Claim 9

A computer implemented method of training a generator neural network for use in generating positions of map items for placement on a virtual map, the method comprising:
receiving one or more training examples, each training example comprising: (i) map data, comprising one or more channels of position information for a region, (ii) a latent vector, and (iii) at least one ground truth placement of map items;
for each training example of the one or more training examples:
generating, with the generator neural network, a generated placement of map items for the training example, comprising processing the map data and the latent vector of the training example;
determining, with a discriminator neural network, an estimated probability of the generated placement of map items being a generated placement of map items or a ground truth placement of map items, comprising processing: (i) the generated placement of map items (ii) the map data of the training example, and (iii) the latent vector of the training example; and
updating parameters of the generator neural network, comprising:
determining an adversarial loss using the estimated probability; and
updating the parameters of the generator neural network in dependence on the adversarial loss.

### Claim 10 (depends on 9)

The method of claim 9, wherein the latent vector of a training example is updated in dependence on the adversarial loss so as to encode a placement configuration of the at least one ground truth placement of map items of the training example.

### Claim 11 (depends on 10)

The method of claim 10, wherein:
the one or more training examples comprise a plurality of training examples; and
one or more of the training examples are cross training examples, a cross training example being derived from a first training example and a second training example, the method further comprising:
replacing the latent vector of the first training example with the latent vector of the second training example.

### Claim 12 (depends on 11)

The method of claim 11, further comprising:
for each cross training example, updating the latent vector of the second training example in dependence on the adversarial loss.

### Claim 13 (depends on 9)

The method of claim 9, wherein updating the parameters of the generator neural network further comprises:
determining a reconstruction loss depending on a comparison between the generated placement of map items and the ground truth placement of map items; and
updating the parameters of the generator neural network parameters in dependence on the reconstruction loss.

### Claim 14 (depends on 13)

The method of claim 13, further comprising:
updating the latent vector of the training example in dependence on the reconstruction loss.

### Claim 15 (depends on 14)

The method of claim 14, wherein the reconstruction loss is determined using a Laplacian pyramid.

### Claim 16 (depends on 9)

The method of claim 9, further comprising:
determining, with a further discriminator neural network, a further estimated probability of the generated placement of map items being a generated placement of map items or a ground truth placement of map items, comprising processing: (i) the generated placement of map items, and (ii) the map data of the training example; and
wherein updating the parameters of the generator neural network further comprises:
determining a further adversarial loss using the further estimated probability; and
updating the parameters of the generator neural network in dependence on the further adversarial loss.

### Claim 17

A computer-readable medium storing instructions, which when executed by a processor, cause the processor to:
receive an input at a generator neural network trained for generating positions of map items for placement on a virtual map, the input comprising:
map data comprising one or more channels of position information for at least a region of the virtual map, said one or more channels including a channel comprising road position information for the region; and
a latent vector encoding a selection of a placement configuration;
generate, with the generator neural network, a probability of placing a map item for each subregion of a plurality of subregions of the region of the virtual map; and
generate placement data of map items for the map data using the probability for each subregion.

### Claim 18 (depends on 17)

The computer-readable medium of claim 17 storing further instructions, which when executed by a processor, cause the processor to:
generate a height for each subregion of a plurality of subregions of the region.

### Claim 19 (depends on 17)

The computer-readable medium of claim 17, wherein the one or more channels further comprises at least one of:
a channel of waterway position information;
a channel of vegetation position information; and
a channel of existing map item position information.

### Claim 20 (depends on 17)

The computer-readable medium of claim 17, wherein the at least one channel of road position information comprises at least one of:
a channel of highway road position information;
a channel of arterial road position information; and
a channel of local road position information.
