3D Map Provisioning With Edge-Cloud Semantic Object Labeling

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current mapping systems for autonomous vehicles rely on cloud processing, which results in stale maps and high computational and bandwidth costs, leading to delays in providing real-time three-dimensional object detection for safe navigation.

Innovation Solution

Implementing a distributed cloud computing environment where autonomous vehicles process and update three-dimensional maps at edge clouds, using point cloud data and two-dimensional images to generate and synchronize real-time maps with semantic information, reducing latency and data size.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If real-time three-dimensional object detection is processed by a central cloud processing environment, then comprehensive mapping can be achieved, but computing power costs and network bandwidth consumption increase significantly

Engineering Contradiction:

Improveobject detection accuracyVSAvoidcomputing power cost

Core Design Contradiction:

Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The patent divides the cloud processing environment into edge clouds distributed near autonomous vehicles. Each edge cloud processes point cloud data and two-dimensional images locally to generate three-dimensional maps, segmenting the centralized computing workload into distributed units that reduce bandwidth consumption and computing costs while maintaining detection accuracy.

Inventive Principle:

Principle #1Segmentation

2Measurement precision

If real-time three-dimensional object detection is processed by a central cloud processing environment, then comprehensive mapping can be achieved, but network bandwidth consumption increases significantly

Engineering Contradiction:

Improveobject detection accuracyVSAvoidnetwork bandwidth consumption

Core Design Contradiction:

Measurement precisionVSLoss of energy

Solution Approach 1:

Inventive Principle:

Principle #1Segmentation

3Reliability

If maps are stored and processed by a central cloud environment, then centralized control can be maintained, but time delays occur in providing real-time maps to autonomous vehicles

Engineering Contradiction:

Improvemap accuracyVSAvoidmap provisioning delay

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The patent introduces edge clouds as intermediary processing nodes between autonomous vehicles and the central cloud environment. These edge clouds receive point cloud data and two-dimensional images, process them locally to generate three-dimensional maps with object detections, and provide real-time updates to vehicles, eliminating the time delays associated with centralized cloud processing while maintaining map accuracy.

Inventive Principle:

Principle #24Intermediary (Mediator)

4Reliability

If comprehensive three-dimensional mapping is provided to autonomous vehicles, then navigation safety is improved, but data size increases leading to higher transmission requirements

Engineering Contradiction:

Improvenavigation safetyVSAvoidmap data size

Core Design Contradiction:

ReliabilityVSQuantity of substance

Solution Approach 1:

The patent extracts only the essential elements needed for safe navigation from comprehensive three-dimensional maps. By using two-dimensional images to detect objects and generate bounding boxes, then mapping these back to three-dimensional space, the system extracts critical object information (position, size, class) without transmitting complete high-resolution three-dimensional map data, thereby reducing data size while maintaining navigation safety.

Inventive Principle:

Principle #2Taking out (Extraction)

Data Source

PatentUS11508118B2Provisioning real-time three-dimensional maps for autonomous vehicles

Publication Date: 2022.11.22 FUJITSU LTD

AI summary

Systems and methods for provisioning real-time three-dimensional maps, including: at a first processing stage: updating a three-dimensional map of an environment based on received point cloud data; extracting three-dimensional proposals of objects of the environment; projecting the three-dimensional proposals onto a two-dimensional image of the environment; generating two-dimensional proposals of the objects of the environment; at a second processing stage: detecting the objects of the environment from the two-dimensional image of the environment; generating two-dimensional bounding boxes of the objects of the environment; matching the two-dimensional bounding box with a particular two-dimensional proposal of the two-dimensional proposals; and labeling, for each two-dimensional proposal that is matched to a two-dimensional bounding box, the three-dimensional proposal corresponding to the two-dimensional proposal with semantic information associated with the two-dimensional bounding box that is matched to the two-dimensional proposal to update the three-dimensional map.