AI-Powered Fleet Repair Efficiency – Automated Parts Recognition & Ordering with EnterpriseDB
Use GenAI to automate manual ordering processes
This system automates vehicle part identification and ordering for fleet mechanics. By using AI-driven image recognition and managing the image embedding through an AI-enabled data layer, the system allows mechanics to take a photo of a vehicle part, automatically identify it, match it to inventory, and place an order with the supplier. This eliminates manual part lookup, reduces vehicle downtime, and streamlines fleet maintenance.
Key Benefits:
- Faster part identification Mechanics don’t need to manually search for parts.
- Automated ordering Reduces human errors in selecting incorrect parts.
- Real-time AI assistance increases efficiency and ensures availability of parts.
- Local OpenShift deployment ensures security, scalability, and data privacy.
- Pre-trained model by Order Management System allows for scale across multiple sites
AI Use Case Workflow
The below steps outline the use case with model training facilitated by OpenShift AI and applications running on a OpenShift cluster deployed in a hybrid cloud environment.:
- Step 1 - Mechanic Assesses defective part
User Roles: End User / Mechanic
The mechanic captures and uploads a photo of a vehicle part using a mobile app.
The image is sent to the Order Management System (OMS) for processing.
- Step 2 - Preparing the Image Corpus with an AI Data Management Layer
EnterpriseDB provides the AI Data Management layer throughout the entire workflow using EDB Postgres AI. This allows for a single solution to manage the AI data end to end by allowing for storage of both image and meta data, auto-generating image embeddings, and by providing semantic search and retrieval of images.
Learn more about how EnterpriseDB and Red Hat work together on the Red Hat Catalog partner page.
User Roles: Data Scientist / AI Engineer
The image data corpus is stored in an external storage array, for instance in an object storage
bucket This storage location is registered in EDB Postgres AI as an AI volume and an automated
semantic AI retriever pipeline is configured for it. This way the image is embedded, stored as a vector and indexed automatically by EDB Postgres AI, which leverages an image embedding encoder model (for instance CLIP) deployed in OpenShift AI.
- Step 3 - Image Processed by a Pretrained AI Model
User Roles: Data Scientist / AI Engineer
The OMS sends a query to a model that was pretrained with the identifying information using IBM Maximo Visual Inspection (MVI).
The model detects and classifies the part, returning the part name, label, and confidence score.
If the part is confidently identified, the OMS proceeds to the order automation and fulfillment in Step 4.
If the confidence is low or the part is not identified, the OMS triggers image Similarity Search outlined below.
Image Similarity Search
User Roles: Machine Learning Engineer / AI Engineer
If the part is not confidently identified by MVI, the OMS triggers an image similarity search in EDB Postgres AI and returns the closest matching images.
If a match is found, the OMS proceeds to Step 4.
If once again no match is found, the part is flagged for manual review and model retraining.
- Step 4 - Automation and Order Fulfillment
User Roles: Database Administrator / Backend Developer
Part Metadata is Retrieved from Relational Database.
The OMS fetches part metadata from EDB Postgres AI.
The metadata includes part number, manufacturer, price, supplier, and stock availability.
Mechanic Confirms the Order
User Roles: End User / Mechanic
The OMS presents the identified part and related data to the mechanic.
The mechanic can confirm the order manually or rely on auto-ordering rules configured by the supplier.
Order is sent to supplier API
User Roles: Backend Developer / API Integration Engineer
Upon confirmation, the OMS places an order through the appropriate supplier API.
Supported supplier APIs include AutoZone, NAPA, RockAuto, or internal fleet inventory systems.
Supplier Ships the Part
User Roles: Supplier / Fulfillment Specialist
The supplier fulfills the order and sends order status and tracking information back to the OMS.
The OMS updates the mechanic with real-time tracking updates.
