2D Distribution Map for Efficient 3D Object Detection

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional systems require significant calculation to detect continuous three-dimensional objects such as side walls, guardrails, and shrubberies on roads, which hinders efficient object recognition and collision avoidance in automotive applications.

Innovation Solution

A processing device with a generating unit, detecting unit, and determining unit that generates two-dimensional distribution information associating lateral and depth direction distances, detects continuous areas in the depth direction, and determines whether these areas represent continuous three-dimensional objects, enabling efficient detection of road boundaries and other objects.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If conventional three-dimensional object detection methods are used, then detection accuracy is maintained, but computational complexity and processing time increase significantly

Engineering Contradiction:
Improvedetection speedVSAvoidcalculation complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The patent transforms three-dimensional object detection into two-dimensional distribution information detection. By projecting 3D spatial coordinates (x, y, z) into a 2D distribution map where pixel positions represent lateral and depth direction distances, the system detects continuous areas in 2D space that correspond to continuous 3D objects. This dimensional reduction significantly decreases computational complexity while preserving detection accuracy.

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

Solution Approach 2:

The patent segments the detection process into distinct functional units: a generating unit that creates two-dimensional distribution information from detection data, a detecting unit that identifies continuous areas in the 2D distribution, and a determining unit that validates whether these areas represent continuous three-dimensional objects. This segmentation allows each unit to perform its specific function efficiently, improving overall detection speed.

Inventive Principle:
Principle #1Segmentation

2Measurement precision

If comprehensive object detection is performed, then recognition accuracy improves, but processing time increases

Engineering Contradiction:
Improveobject recognition accuracyVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

By representing three-dimensional object distribution as two-dimensional information where the horizontal axis represents lateral direction distance and the vertical axis represents depth direction distance, the patent enables parallel processing of spatial relationships. Continuous areas in this 2D distribution can be detected simultaneously across the entire field of view, maintaining high recognition accuracy while reducing processing time compared to sequential 3D analysis.

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

Data Source

PatentUS10748014B2Processing device, object recognition apparatus, device control system, processing method, and computer-readable recording medium
Publication Date: 2020.08.18 RICOH CO LTD
  • US10748014B2 patent drawing
  • US10748014B2 patent drawing
  • US10748014B2 patent drawing

AI summary

According to an embodiment, a processing device includes a generating unit, a detecting unit, and a determining unit. The generating unit is configured to generate two-dimensional distribution information of an object, the two-dimensional distribution information associating between at least a lateral direction distance and a depth direction distance of the object. The detecting unit is configured to detect a continuous area having continuity in a depth direction in the two-dimensional distribution information. The determining unit is configured to determine whether the continuous area represents a detection target.