Stereo Camera U-Map Depth Separation for Collision Avoidance

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current collision avoidance systems face challenges in accurately detecting and separating objects at long distances due to deteriorating range resolution, often combining multiple objects into one, leading to inaccurate detection.

Innovation Solution

The implementation of a stereo camera system with a CPU-controlled image processing device generates a U-map and performs depth separation using one-dimensional histograms to accurately separate objects based on disparity values, enhancing resolution and detection accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Length of stationary object

If stereo vision or millimeter-wave radar is used to detect objects at long distance, then detection range is extended, but range resolution deteriorates causing multiple objects to be combined into one

Engineering Contradiction:
Improvedetection rangeVSAvoidrange resolution
Core Design Contradiction:
Length of stationary objectVSMeasurement precision

Solution Approach 1:

The patent segments objects that were previously detected as a single combined object by using depth information from U-maps to separate them into individual objects based on their spatial positions and depth values

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces depth as an additional dimension by generating U-maps from stereo vision data, transforming 2D image data into 3D spatial information to resolve objects that overlap in 2D but are separated in depth

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Speed

If object detection is performed using conventional methods, then detection speed is maintained, but accuracy deteriorates when objects are at long distance

Engineering Contradiction:
Improvedetection speedVSAvoidobject detection accuracy
Core Design Contradiction:
SpeedVSMeasurement precision

Solution Approach 1:

The patent performs preliminary depth calculation and U-map generation before final object detection, preparing depth information in advance to improve subsequent object separation and detection accuracy

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent introduces U-maps as an intermediary data structure that bridges stereo vision data and object detection, enabling accurate depth-based object separation without compromising detection speed

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentEP3324336B1Information processing device, information processing method, and carrier means
Publication Date: 2024.06.26 RICOH CO LTD
  • EP3324336B1 patent drawingFigure 1~2
  • EP3324336B1 patent drawingFigure 3
  • EP3324336B1 patent drawingFigure 4

AI summary

The information processing device (30) and the information processing method include generating two-dimensional distribution data in which a position of an object in a horizontal direction and a position of the object in a depth direction are associated with a frequency value, detecting the object, based on the two-dimensional distribution data, integrating the frequency value according to a position in the depth direction to generate amounts of characteristics, specifying a to-be-separated position based on the amounts of characteristics, and separating the object at the to-be-separated position. The carrier means carries computer readable code for controlling a computer to carry out the information processing method.