1. Project Description, Objectives, and Priorities
The QUICKPICK project arises from a critical need in the retail sector and small local shops: the removal of physical and operational barriers in the purchasing process. For many merchants, manual inventory management and checkout wait times limit their competitiveness against e-commerce. The overall goal of the project is to develop an automated shopping system based on Computer Vision and IoT that transforms the physical shopping experience into a fluid, frictionless, and fully digitalized process.
The result is conceived as a technological ecosystem that enables automatic product detection, real-time virtual cart generation, and centralized management of multiple points of sale, all integrated into a high-performance cloud architecture.
Main Objective: To develop a disruptive automated shopping solution for retail that, through optical recognition algorithms and smart sensors, eliminates the need for manual product scanning. The platform aims to offer a "Grab & Go" experience, allowing customers to shop autonomously via an intuitive mobile app synchronized with the physical store.
Specific Objectives:
-
Scalable Cloud Architecture: Build a robust infrastructure on AWS using services such as EKS (Elastic Kubernetes Service) and S3, optimized for processing real-time data streams from sensors and cameras.
-
Computer Vision and Product Recognition: Develop and train Machine Learning and computer vision models to identify products accurately, minimizing errors under varying lighting conditions and store layouts.
-
Multi-platform Frontend Development: Create a mobile application using Flutter/Dart to serve as the primary user interface, enabling virtual cart visualization and secure payments.
-
Privacy and Ethical Development: Prioritize GDPR compliance by pivoting from the initial facial recognition design toward QR code identification and object tracking systems to ensure maximum protection of biometric data.
-
Inventory Management Optimization: Integrate tools for retailers to monitor KPIs, reduce perishable waste, and automate stock replenishment.
-
Testing and Validation Cycle: Execute a rigorous program of unit and system tests to validate product detection reliability and security within integrated payment gateways.
2. Geographic or Thematic Scope of Execution
This project is at the forefront of Artificial Intelligence applied to Retail. It represents a significant step toward more efficient and sustainable business models, with the potential to revitalize physical commerce through the adoption of technologies previously only accessible to large international corporations.
3. Total EU Aid Assigned, Budget, or Total Investment
The total approved budget for the QUICKPICK project amounts to €196,337.00. The CDTI has granted a loan-based aid of €166,886.00 to the company Macrotest Informática S.L. This aid is co-financed by the European Regional Development Fund (ERDF), under the slogan "A way to make Europe."
4. Duration or Execution Period
The project began with its presentation on 12/19/2023. Following an extension request motivated by the complexity of the technical integrations in Milestone 2, the execution and justification period has been extended until 11/30/2025.
5. European Funds Brand Graphic Elements
Project Results
Implementation of Project Management Methodology
Application of an Agile framework for monitoring software development and hardware integration. During the first milestone, Jira was used for task traceability and change control. Technical risk management protocols have been established, focusing specifically on interoperability between IoT sensors and the cloud backend.
Definition of Requirements
Exhaustive analysis of retail sector needs to define the functional requirements of the vision system and the mobile app. The network architecture was designed to minimize store blind spots, and User Experience (UX) flows were defined. A key outcome was the technical report justifying the replacement of facial recognition with less intrusive technologies (QR/Object Vision) to comply with GDPR.
-
Deliverables:
-
System requirements technical report.
-
Architecture design document and UI/UX prototyping.
-
Data Protection Impact Assessment (DPIA/EIPD).
Platform and App Development
Construction of the backend on AWS and development of the QuickPick mobile app (Android/iOS). The app allows for instant synchronization with the store, virtual cart management, and integration of electronic payment systems. Data encryption has been implemented to ensure transactions and user privacy.
-
Deliverables:
-
QuickPick application (functional prototype version).
-
Source code repositories (Frontend and Backend).
-
Cloud infrastructure configuration (AWS).
Artificial Intelligence Integration
Development of the core computer vision engine for real-time product recognition. Deep learning models have been adapted to detect when items are taken or returned, and a personalized recommendation engine based on purchase history has been designed to improve the customer experience.
-
Deliverables:
-
Technical documentation of computer vision models.
-
Recommendation algorithms and business intelligence.
-
Model accuracy validation report.
Platform Testing
Execution of load tests on the cloud infrastructure and validation of optical recognition accuracy. Integration tests were performed between hardware (cameras) and software (App) to ensure minimum response times. The testing plan includes verifying payment gateway security and system robustness during network outages.
-
Deliverables:
-
Test case report and bug fixes.
-
Validation plan with real users (Beta Testers).
-
Internal security certification and payment protocols.
