Image-Based Delivery Location Scoring for Theft and Access Risk

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

The increasing number of online orders for package delivery has led to issues such as packages being stolen or delivered to incorrect addresses, and deliverers facing difficulties in locating or accessing delivery locations due to poor signage, foliage, fences, or aggressive pets, resulting in poor customer experience and increased costs.

Innovation Solution

A system that analyzes images of delivery locations using classifiers to determine scene attributes and correlate them with delivery attributes, generating scores to provide delivery instructions that help deliverers navigate challenges such as theft risk, access issues, and parking difficulties.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If the number of online orders for package delivery increases, then delivery volume increases, but the risk of package theft and incorrect deliveries increases

Engineering Contradiction:

Improvedelivery volumeVSAvoiddelivery accuracy

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The system performs preliminary analysis of delivery locations using images and scene attributes before delivery occurs. Delivery instructions are generated in advance based on predicted delivery attributes, allowing deliverers to prepare appropriately and reducing errors during the actual delivery process

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system introduces an intermediary layer between the order system and the deliverer. This intermediary analyzes delivery locations using scene attributes from images and generates predictive delivery instructions, acting as a mediator that improves delivery accuracy without increasing delivery volume capacity

Inventive Principle:

Principle #24Intermediary (Mediator)

2Reliability

If deliverers are provided with more delivery location information, then delivery accuracy improves, but the time and resources required to assess each location increase

Engineering Contradiction:

Improvedelivery accuracyVSAvoidlocation assessment time

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The system uses image copies or representations of delivery locations to extract scene attributes and predict delivery attributes. Instead of requiring deliverers to physically assess each location, the system creates a digital model from images and generates instructions based on this copy, significantly reducing assessment time while maintaining accuracy

Inventive Principle:

Principle #26Copying

Solution Approach 2:

The system transforms physical delivery location characteristics into measurable scene attributes from images (such as foliage density, signage visibility, access point characteristics). By converting physical assessment into image parameter analysis, the system reduces the time required while improving consistency and accuracy of delivery location evaluation

Inventive Principle:

Principle #35Parameter changes

3Ease of operation

If deliverers manually assess each delivery location, then delivery instructions can be customized, but the process becomes complex and error-prone

Engineering Contradiction:

Improvedelivery instruction customizationVSAvoidassessment process complexity

Core Design Contradiction:

Ease of operationVSDevice complexity

Solution Approach 1:

The system enables self-service by automatically analyzing delivery location images and generating customized delivery instructions without requiring manual deliverer assessment. The system serves itself by using scene attributes to predict delivery attributes and generate appropriate instructions, reducing operational complexity while maintaining customization

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The system replaces the mechanical process of manual deliverer assessment with an automated image analysis and machine learning system. Scene attribute extraction and delivery attribute prediction algorithms substitute for human judgment, reducing complexity and errors while maintaining the ability to provide customized delivery instructions

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

4Reliability

If deliverers are given detailed delivery instructions, then delivery success rate improves, but the cost of generating and processing instructions increases

Engineering Contradiction:

Improvedelivery success rateVSAvoidinstruction processing cost

Core Design Contradiction:

ReliabilityVSLoss of energy

Solution Approach 1:

The system applies partial action by generating delivery instructions only for locations where scene attributes indicate potential delivery challenges. For straightforward locations with favorable scene attributes, minimal or no additional instructions are provided, reducing processing costs while maintaining high delivery success rates where they are most needed

Inventive Principle:

Principle #16Partial or excessive action

Data Source

PatentUS11151509B1Image-based scene attribute and delivery attribute determination

Publication Date: 2021.10.19 AMAZON TECH INC

AI summary

Images that depict delivery locations associated with a customer may be captured or obtained. One or more classifiers may be utilized to detect and/or identify scene attributes that are depicted in the images and that are associated with the delivery location. One or more correlations between the scene attributes and delivery attributes associated with the delivery location may be determined and used to generate one or more scores for the delivery location. The correlation(s) and/or the score(s) may be utilized to determine delivery instructions that facilitate the delivery of items and/or packages to the delivery location.