Overview

I competed as a global finalist in Siemens #KicksForEdge, a 72-hour hackathon held at Munich Urban Colab. The event was part of IPCEI-CIS, advancing interoperable cloud–edge technologies.

Global Finalist Munich Urban Colab IPCEI-CIS Industrial Edge
Partner
Element Logic Germany
Track
Edge log analytics
Problem
Conveyor misdirection

What we built

Hidden faults were buried inside raw PLC logs even when the system reported “OK”. Over 72 hours, we built an edge-based log analytics pipeline to convert unstructured logs into actionable technician insights. Warehouse technicians manually sift through millions of log lines across dozens of PLCs. Unflagged errors cause conveyor arms to misroute packages without triggering alarms. These “silent” failures accumulate into enormous financial loss. Our research showed that Germany alone reported 44,406 operational complaints in 2024 related to delayed/misdirected parcels. (Source included in slide deck)

Hackathon tracks (3)
  • Challenge 1 — Edge-enabled data analysis (Element Logic Germany)
  • Challenge 2 — Hydropower predictive maintenance (LEW Wasserkraft)
  • Challenge 3 — Robot QC from sensor data (HOLZ automation)
Deployment
Siemens SIMATIC IPC BX-39A
Industrial Edge device
Example PLC log excerpt
synthetic
[12:14:03.112] 'Logger 3 "UNIT-A" (192.168.1.50:5000)' connected!
[12:14:03.129] [COM RECV] RECV STATUS 16#7001
[12:14:03.145] [COM FILL] 01|CV|000101|N|FL|000000//0000|...
[12:14:03.169] [COM RECV] CV|02|000000|A|5|000002//...|SR|...
[12:14:03.332] [COM RECV] CV|02|000000|R|30|000000//...|SR|...
[12:14:03.387] FIFO warning: buffer usage 85%
Synthetic data only. Used to demonstrate parsing and fault surfacing.

Our Solution

We built a full pipeline using Node-RED → OpenSearch → Dashboards.

  • Automated log ingestion for text/ZIP PLC logs
  • Custom parser that cleans, extracts, and formats raw strings into JSON
  • Device, alarm, and routing analysis to identify common failure clusters
  • Interactive dashboards for non-engineers to visualize trends instantly
Impact
Reduced troubleshooting from 7 days to 10 minutes.
01 Log Uploader
Log uploader UI

Upload PLC logs (TXT/ZIP) for processing.

02 Node-RED Parser
Node-RED flow

Clean + parse raw strings into structured JSON.

03 OpenSearch Dashboard
OpenSearch dashboard

Searchable insights + trend views for technicians.

Future Vision

With a unified dataset, the next step is applying Edge AI models to detect anomalies before they become misdirection events. A chatbot interface over OpenSearch gives technicians instant, conversational access to system history and root-cause explanations.

Edge AI anomaly detection OpenSearch RAG-style search Technician-friendly UX
Goal
Proactive alerts → fewer stoppages
Interface
Chat-based root-cause workflow
Where it runs
Near machines (Industrial Edge)
Prototype UI AI Chatbot Assistant
AI Chatbot Assistant mockup
Concept mockup (synthetic / non-confidential) for demonstrating the workflow.

Slide deck

Pictures

Contact Me

senrinakamura60@gmail.com

+1 781 539-3225

Resume