MADDOX solves the problem that even the best-maintained knowledge database is useless if, due to a lack of experience, you don’t know what to look for when a malfunction occurs in production. Specifically, the correct term for the affected machine parts, for example, is often unknown or ambiguous, or there are dozens of possible causes for the symptom or error code specified in the machine error message. MADDOX’s self-learning, AI-based search algorithm recognizes characteristic patterns and correlations in the fault descriptions and the associated solutions, enabling it to identify relevant information and possible causes of faults even with vague or incomplete search queries.
Most of our customers want to host MADDOX on premise, either on a hardware server or virtualized in their own data center. This means that all data remains with the customer.
MADDOX runs as a Docker container, preferably on Linux. Our system integration team takes care of setup, system monitoring, and updates in close coordination with the customer’s IT department and using remote maintenance access. This allows us to combine the simple, maintenance-free availability of a cloud solution with the security of knowing that customer data never leaves the company network.
If you would prefer to use MADDOX as a real cloud application for better scalability and greater reliability, we offer our own GDPR-compliant cloud hosting at Hetzner Online GmbH in the data centers in Falkenstein and Nuremberg. Alternatively, you can of course also use your own corporate cloud solutions, such as virtual private clouds (VPCs) from AWS & Co.
Our integration experts will work with you in detail to determine your preferred hosting solution as part of a pilot project and then take care of the implementation.
Our simple principle here is that the data belongs to the person who generated it.
In most cases, this is the machine operator, who is our customer, where the knowledge entries and training data on malfunctions are created by their own employees and machines during ongoing production. Machine manufacturers, who are also among our customers, generate training data in their own technical centers or during the commissioning of machines at the customer’s site.
In order for property rights to be exercised in practice, it is important that the relevant data is accessible. MADDOX offers the option of exporting complete knowledge entries as PDF files, exporting photos and videos individually, or accessing all data via an open REST API interface.
Peerox reserves the right to use training data to test new or optimized machine learning algorithms. Experience shows that real machine data often contains unexpected and difficult-to-simulate effects such as gaps, outliers, incorrect implementations of specifications, etc. Therefore, testing with real data is essential for the continuous development of our algorithms for the benefit of our customers in order to avoid “unpleasant surprises” during updates. To this end, we roll out the new algorithm as a Docker container on selected customer servers, train it with the data available there, compare the results with the existing algorithm, and then delete the Docker container again. In this process, customer data does not leave the customer server either directly or indirectly, for example as a trained machine learning model. For Peerox, the only relevant information is whether the new algorithm works stably and has delivered better results than the old one.
MADDOX is a server-based application with a web front end. This means MADDOX can run on virtually any device that can display a web page and connect to the server (laptop/PC, tablet, smartphone, touch panels, etc.). The web front end is also delivered as a progressive web app (PWA) that can be installed without admin rights on iOS, Windows, Android, and Linux. A shortcut is created on the home screen that opens MADDOX in a special window of the system browser without any controls, which feels like a native app to users but offers all the advantages of a web application (security, system independence, etc.).
The MADDOX authentication system is based on the open source access and identity management (IDM) system Keycloak, which can be connected to existing user management systems via LDAP, OpenID Connect, Kerberos, SAML 2.0, and other technologies. This enables single sign-on (SSO) and central user management in your existing IDM.
MADDOX has an SAP interface that allows high-quality SAP messages to be generated from the data available in MADDOX with minimal effort on the part of the user. It is also possible to read relevant production data from HANA databases.
If you would like to use MADDOX on premise (see section “How is MADDOX hosted?”), our system integration team will work with your IT experts to design and implement a system architecture for integrating MADDOX into your IT landscape as part of a pilot project. We can implement even the most complex solutions with distributed installations, jump servers between different network partitions separated by firewalls, or hosting on your specially hardened operating systems.
MADDOX natively supports the modern protocols OPC UA and MQTT. In addition, we follow the approach of not reinventing the wheel and therefore use the Robogate product from Robotron Datenbank-Software GmbH as a gateway solution. This supports most common interfaces such as Modbus TCP/RTU, RFC1006 (Siemens S7), and other protocols. Robogate runs as a Docker container directly on the MADDOX server and is configured and managed by us, so there is no additional effort on the part of the customer. For EtherCAT, EtherNet/IP, ProfiNET, and non-Ethernet-based hardware interfaces such as ProfiBus or “exotic” data sources (e.g., dew point sensors), we use products from the RevolutionPi series from KUNBUS GmbH, which we also set up completely so that you don’t have to worry about anything.
MADDOX’s modular service architecture also allows us to easily implement customer-specific native interfaces. Solutions already implemented include SQL, InfluxDB, Apache Kafka, and proprietary TCP/IP protocols.
In short: machine data interfaces are diverse, but in our more than five-year company history, the connection has never failed for technical reasons. In a pilot project, our interface experts work with you to develop a solution for your machinery.
MADDOX is a server application, which means that a connection must exist between the end device (e.g., tablet) and the MADDOX server. You therefore need Wi-Fi in the production area to use it on mobile devices. However, if you have opted for on-premise hosting in your intranet (see section “How is MADDOX hosted?”), no Internet connection is required.
