# US 2019/0294934 A1 · Method, apparatus, and system for dynamic adaptation of an in-vehicle feature detector

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

- Number: US 2019/0294934 A1
- Document ID: US20190294934A1
- Status: Published application
- Kind: A1
- Application number: 15/933,063
- Filed: 2018-03-22
- Published: 2019-09-26
- Earliest claimed: 2018-03-22
- Assignees: HERE Global B.V.
- Applicants: Here Global B.V.
- Inventors: Vladimir SHESTAK; Stephen O'HARA; Nicholas DRONEN
- Web page: https://patentgrep.com/patent/US20190294934A1

## CPC

- G06V 20/56: exterior to a vehicle by using sensors mounted on the vehicle
- G06V 10/82: using neural networks
- G06V 10/87: {using selection of the recognition techniques, e.g. of a classifier in a multiple classifier system}
- G06V 10/95: {structured as a network, e.g. client-server architectures}
- G06V 20/588: {Recognition of the road, e.g. of lane markings; Recognition of the vehicle driving pattern in relation to the road}
- B60W 30/08: {Active safety systems} predicting or avoiding probable or impending collision {or attempting to minimise its consequences}
- B60W 2420/403: Image sensing, e.g. optical camera
- B60W 2552/00: Input parameters relating to infrastructure
- B60W 2552/53: Road markings, e.g. lane marker or crosswalk
- G06N 20/00: Machine learning
- H04L 67/12: specially adapted for proprietary or special-purpose networking environments, e.g. medical networks, sensor networks, networks in vehicles or remote metering networks
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- 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
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- H04L: TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION (arrangements common to telegraphic and telephonic communication H04M)

## Abstract

An approach is provided for dynamic adaptation of an in-vehicle feature detector. The approach involves embedding a feature detection model, precomputed weights for the feature detection model, or a combination thereof in a data layer of map data representing a geographic area from which a training data set was collected to generate the feature detection model, the precomputed weights, or a combination thereof. The approach also involves deploying the feature detection model, the precomputed weights, or a combination thereof to adapt an in-vehicle feature detector based on determining that the in-vehicle feature detector is in the geographic area, plans to travel in the geographic area, or a combination thereof. The in-vehicle feature detector can then use the feature detection model, the precomputed weights, or a combination thereof to process sensor data collected while in the geographic area to detect one or more features.

## Claims

### Claim 1

A computer-implemented method comprising:
embedding, by a processor, a feature detection model, precomputed weights for the feature detection model, or a combination thereof in a data layer of map data representing a geographic area from which a training data set was collected to generate the feature detection model, the precomputed weights, or a combination thereof; and
deploying the feature detection model, the precomputed weights, or a combination thereof to adapt an in-vehicle feature detector based on determining that the in-vehicle feature detector is in the geographic area, plans to travel in the geographic area, or a combination thereof.

### Claim 2 (depends on 1)

The method of claim 1, wherein the in-vehicle feature detector uses the feature detection model, the precomputed weights, or a combination thereof to process sensor data collected while in the geographic area to detect one or more features.

### Claim 3 (depends on 1)

The method of claim 1, wherein the map data is organized into a tile-based structure, and wherein the geographic area corresponds to one map tile of the tile-based structure.

### Claim 4 (depends on 1)

The method of claim 1, further comprising:
embedding another feature detection model, other precomputed weights for the another feature detection model, or a combination thereof respectively in each map tile of the tile-based structure,
wherein the another feature detection model, the other precomputed weights, or a combination are based on a respective training data set collected from said each map tile.

### Claim 5 (depends on 1)

The method of claim 1, wherein the feature detection model, the precomputed weights, or a combination there are further based on one or more operational conditions of the in-vehicle feature detector, and wherein the deploying of the feature detection model, the precomputed weights, or a combination thereof is further based on determining that the in-vehicle feature detector is operating or expected to operate under the one or more operational conditions.

### Claim 6 (depends on 1)

The method of claim 1, wherein one or more detection classes of the feature detection model are further based on the training data set collected from the geographic area.

### Claim 7 (depends on 1)

The method of claim 1, wherein the feature detection model, the precomputed weights, or a combination thereof are embedded in a data layer of the map data.

### Claim 8 (depends on 1)

The method of claim 1, wherein the feature detection model, the precomputed weights, or a combination thereof are determined in an offline mode; and wherein the deploying of the feature detection model, the precomputed weights, or a combination thereof is performed dynamically in an online mode.

### Claim 9 (depends on 1)

The method of claim 1, wherein the in-vehicle feature detector is part of computer vision system of a vehicle.

### Claim 10

An apparatus comprising:
at least one processor; and
at least one memory including computer program code for one or more programs, the at least one memory and the computer program code configured to, with the at least one processor, cause the apparatus to perform at least the following,
retrieve a map tile from a geographic database, wherein the map tile includes a data layer storing a feature detection model, precomputed weights for the feature detection model, or a combination thereof; and wherein the feature detection model, the precomputed weights, or a combination thereof are generated using a training data set collected from a geographic area represented by the map tile; and
adapt an in-vehicle feature detector with the feature detection model, the precomputed weights, or a combination thereof based on determining that the in-vehicle feature detector is in or expected to be in the geographic area,
wherein the adapted in-vehicle feature detector is used to process sensor data collected from the geographic area to detect one or more features.

### Claim 11 (depends on 10)

The apparatus of claim 10, wherein the retrieving of the map tile comprises pre-fetching the map tile based on a planned location, a planned route, or a combination thereof of a vehicle equipped with the in-vehicle feature detector.

### Claim 12 (depends on 10)

The apparatus of claim 10, wherein the apparatus is caused to:
store the feature detection model, the precomputed weights, or a combination thereof in the at least one memory of the apparatus, another memory of the in-vehicle feature detector, or a combination thereof.

### Claim 13 (depends on 12)

The apparatus of claim 12, wherein the stored feature detection model, the stored precomputed weights, or a combination thereof are used to dynamically adjust the in-vehicle feature detector when the in-vehicle feature detector is in or expected to be in the geographic area.

### Claim 14 (depends on 10)

The apparatus of claim 10, wherein the map tile is among a plurality of map tiles of the geographic database, wherein each of the plurality of map tiles includes a respective data layer comprising a respective feature detection model, respective precomputed weights of the respective feature detection model, or a combination thereof.

### Claim 15 (depends on 10)

The apparatus of claim 10, wherein the feature detection model, the precomputed weights, or a combination thereof are generated with respect to one or more operational conditions, and wherein the in-vehicle feature detector is configured with the feature detection model, the precomputed weights, or a combination thereof when the in-vehicle feature detector is operating or expected to operate under the one or more operational conditions.

### Claim 16

A non-transitory computer-readable storage medium carrying one or more sequences of one or more instructions which, when executed by one or more processors, cause an apparatus to perform:
generating a map layer for one or more map tiles of a tile-based geographic database, the map layer storing a feature detection model, precomputed weights for the feature detection model, or a combination thereof respectively for the one or more map tiles,
wherein the feature detection model, the precomputed weights, or a combination thereof are generated from a training data set respectively from the one or more map tiles; and
wherein the feature detection model, the precomputed weights, or a combination thereof are dynamically deployed to an in-vehicle feature detector in response to a location-based request for the one or more map tiles.

### Claim 17 (depends on 16)

The non-transitory computer-readable storage medium of claim 16, wherein the precomputed weights are calculated during an offline training of the feature model using the training data set.

### Claim 18 (depends on 17)

The non-transitory computer-readable storage medium of claim 17, wherein the feature detection model is a neural network, and wherein the precomputed weights are associated with a plurality of connections among a plurality of interconnected processing elements of the neural network.

### Claim 19 (depends on 16)

The non-transitory computer-readable storage medium of claim 16, wherein one or more detection classes of the feature detection model are further based on the training data set collected from the geographic area.

### Claim 20 (depends on 16)

The non-transitory computer-readable storage medium of claim 16, wherein the in-vehicle feature detector uses the feature detection model, the precomputed weights, or a combination thereof corresponding to a map tile in which the in-vehicle feature detector is located or expected to be located.
