Obstacle Detection Using Ground Surface Shape Estimation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing obstacle detection systems inaccurately identify road surfaces as obstacles due to low sensor accuracy, leading to unreliable obstacle detection, especially when the road surface shape estimation is inaccurate.

Innovation Solution

An information processing apparatus that includes a position acquisition unit, a shape acquisition unit, and a calculation unit, which acquires three-dimensional position information of detection points, estimates the ground surface shape, and calculates the existence probability of obstacles based on the distance of detection points from the ground surface, thereby distinguishing between travelable regions and obstacles.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If road surface shape estimation is performed using a single sensor, then device complexity is reduced, but measurement precision deteriorates leading to erroneous obstacle detection

Engineering Contradiction:
Improvesensor quantityVSAvoidroad surface shape estimation accuracy
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The patent combines data from a laser sensor and a camera to estimate road surface shape. The laser sensor provides distance information while the camera provides visual information, and their fusion enables accurate road surface estimation without requiring additional specialized sensors, thus resolving the contradiction between device complexity and measurement precision.

Inventive Principle:
Principle #5Merging (Combining)

2Measurement precision

If detection threshold is lowered to improve obstacle detection sensitivity, then detection precision improves, but false positive rate increases due to road surface misclassification

Engineering Contradiction:
Improveobstacle detection accuracyVSAvoidfalse positive rate
Core Design Contradiction:
Measurement precisionVSReliability

Solution Approach 1:

The patent introduces road surface shape estimation as an intermediary step between raw sensor data and obstacle detection. By first estimating the road surface shape and then comparing detection points against this estimated surface, the system can accurately distinguish between road surface points and actual obstacles, enabling low detection thresholds without increasing false positives.

Inventive Principle:
Principle #24Intermediary (Mediator)

3Adaptability or versatility

If another sensor is added to detect obstacles in directions where points are removed, then obstacle detection coverage is improved, but device complexity and cost increase

Engineering Contradiction:
Improveobstacle detection coverageVSAvoidsensor system complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent makes the existing sensor system multi-functional by using the same laser sensor and camera for both road surface estimation and obstacle detection. The processed road surface shape information serves dual purposes: defining the drivable area and providing a reference for obstacle detection in all directions, eliminating the need for additional specialized sensors.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Data Source

PatentEP3346237B1Information processing apparatus, information processing method, and computer-readable medium for obstacle detection
Publication Date: 2023.11.15 KK TOSHIBA
  • EP3346237B1 patent drawingFigure 1~2
  • EP3346237B1 patent drawingFigure 3
  • EP3346237B1 patent drawingFigure 4A~4B

AI summary

According to an arrangement, an information processing apparatus(20) includes a position acquisition unit(20C), a shape acquisition unit(20D) and a calculation unit(20E). The position acquisition unit(20C) acquires three-dimensional position information of a detection point. The shape acquisition unit(20D) acquires target surface shape information indicating a shape of a target surface. The calculation unit(20E) calculates state information of an object according to a first distance of the detection point from the target surface on the basis of the three-dimensional position information and the target surface shape information.