Inspiration
Airline catering logistics is a symphony of precision performed under immense time pressure. As we observed this complex ballet, we realized that the core issue wasn’t a lack of effort—it was the reliance on manual processes that are inherently slow, error-prone, and difficult to standardize. We saw a clear opportunity: to transform the value chain through digitalization and applied intelligence.
Our inspiration was to build not just a solution to a problem, but a complete ecosystem—a bridge to the future of “Pick & Pack,” a system that empowers employees, optimizes resources, and guarantees perfection on every flight.
What it does
We developed a comprehensive and modular platform that tackles key pain points in airline catering logistics, merging data intelligence with real-world execution. Our solution is built on four interconnected pillars:
Mobile Command Center:
A cross-platform mobile app (iOS/Android) that serves as the main interface for employees. It digitizes manual tasks by scanning QR codes on each flight cart bucket, providing visual and intuitive control over inventory, product status, and assembly tasks.
Productivity Optimizer:
A sophisticated mathematical model that standardizes and predicts cart assembly times. Instead of relying on costly AI models, our approach offers precise time forecasts for scheduling, performance benchmarking, and consistent workforce optimization.
Smart Inventory Brain:
A dual-engine system that addresses two of the most complex challenges:
Consumption Prediction: Using a Weighted Moving Average (WMA) model, we predict product demand for specific flights, generating recommendations to reduce waste and prevent overstocking.
Automated Quality Control: Through Markov Chains, we built a decision system for managing alcohol bottles, determining whether to reuse or discard them based on data sequences—removing subjectivity and ensuring compliance.
Automated Eye:
A computer vision system powered by a YOLO model for automatic counting of objects such as cans and bottles inside boxes. Integrated into a conceptual workstation, it scans the box’s QR code, identifies contents, validates quantities against forecasts, and detects errors in real time—drastically accelerating verification and instant data updates.
Together, these components create a cohesive ecosystem that shifts operations from reactive to proactive, from manual to intelligent.
How we built it
Our solution is the result of diverse skills and technologies working in synergy, built upon a robust and scalable architecture.
Mobile App (Frontend):
Developed in React Native with JavaScript, allowing us to deploy a single codebase natively on both iOS and Android, ensuring accessibility and fast implementation.
Infrastructure (Backend):
The backbone of our system is a reliable, reproducible API encapsulated in a Docker container—ensuring 24/7 operation across any OS in an isolated, controlled environment.
Intelligence and Data Models:
- For productivity optimization, we designed a custom mathematical model that achieves high accuracy without the overhead of traditional machine learning.
- For inventory management, we implemented WMA and Markov Chain models in Python to handle prediction and sequential decision-making effectively.
- For rapid prototyping, we used SQL to validate our database concepts before integrating with the main API.
Computer Vision and Automation:
The object-counting system was built in Python using a YOLO object detection model. Here lies one of our biggest innovations: we trained YOLO on photorealistic synthetic data generated with SolidWorks. We designed and rendered 3D boxes and products, producing perfectly labeled training images. This unique blend of mechatronic engineering and data science allowed us to overcome the lack of real-world data. Final labeling was managed with CVAT, hosted entirely on Google Cloud.
Challenges we ran into
The journey to our final solution was full of challenges that pushed us to adapt, learn, and collaborate intensively.
AI Trap:
At first, we tried conventional AI models for consumption prediction but found they failed to meet our accuracy and efficiency standards. This forced us to rethink our approach—pivoting toward mathematical models like WMA and Markov Chains, a crucial decision for our project’s success.
Parallel Development:
Building the mobile app and the API simultaneously created significant coordination challenges. Continuous communication was essential to refine API endpoints and ensure perfect integration.
Crossing Disciplines:
Our team brought together diverse expertise—an advantage, but also a challenge. A mechatronics engineer had to learn database management under pressure, while a mobile developer faced integrating complex 3D models in JavaScript with no prior experience. Overcoming these hurdles required trust, mutual support, and a strong willingness to learn.
Rendering Dilemma:
Generating synthetic data in SolidWorks was innovative but computationally expensive. With time against us, we had to strategically choose which elements were worth rendering in high detail, optimizing our most valuable resource: time.
Accomplishments that we're proud of...
A Solution with Immediate Impact:
More than a prototype, we built a functional, cohesive product capable of generating tangible change from day one. The mobile app is ready to deploy, digitizing processes and improving efficiency instantly.
Innovation at the Intersection of Fields:
Our greatest pride is merging disciplines—using mechatronic engineering and 3D design with SolidWorks to generate synthetic data and solve a computer vision problem. It proves that the best solutions often come from thinking beyond traditional programming boundaries.
The Elegance of Simplicity:
We’re proud to have recognized the limits of complex AI models and chosen simpler, more elegant mathematical solutions. Our productivity model—predicting 4.31 minutes per cart—shows that the right tool, not necessarily the newest, delivers the best results.
Resilience and Collaboration:
Completing such a complex ecosystem, from backend to computer vision, under time constraints, is an achievement we value deeply. Every member overcame personal technical and knowledge challenges, showcasing incredible adaptability and teamwork.
What we learned...
The Power of the Unexpected:
The most valuable lesson was that no skill is irrelevant. What began as a coding competition was unexpectedly enriched by mechanical design expertise in SolidWorks. We learned that true innovation happens when diverse perspectives and tools from different fields combine to solve a problem.
Mathematical Elegance Over Complexity:
We demonstrated the immense power of fundamental mathematical models. In a world obsessed with machine learning, we learned that applying solutions like WMA, Markov Chains, or simulation models can be more cost-effective, precise, and efficient for certain business challenges.
Synergy of Ideas:
We learned that combining different perspectives multiplies the final result. By sharing our viewpoints, data science concepts blended with robotics and automation ideas—transforming our project from a simple app into an ambitious, forward-looking proposal.
What's next for GateGroup SmartIntelligence - Team oVerSimplified
We see this project not as an endpoint, but as the foundation of a much larger platform. Our vision unfolds in two strategic phases:
Scaling and Refining the Software:
The next step is to scale the current platform—expanding predictive and optimization capabilities while enriching the mobile app with real time user feedback-driven features, consolidating it as the central operations tool and build simulation capabilities for scenario planning and strategy testing.
Integrating with Physical Automation:
Our mid-term vision is to bridge the gap between digital and physical. We plan to integrate our software ecosystem with robotics hardware. The computer vision system is just the first step—the ultimate goal is to create truly automated “Pick & Pack” stations, where robotic arms, guided by our platform’s intelligence, perform assembly tasks with unprecedented efficiency and precision.
Team oVerSimplified: Where analytical elegance meets operational excellence.
Built With
- cvat
- expo.io
- flask
- javascript
- opencv
- python
- react-native
- solidworks
- sql
Log in or sign up for Devpost to join the conversation.