Control Points | Vision MES

Control Points Explained

/01

Machine State Monitoring

An AI model monitors the machine’s tower-light and classifies its colour states as RUN / IDLE / FAULT. Every transition triggers an automatic event ticket and links a video clip, building a precise timeline of machine availability and stoppages.

Outputs:

📊 Second-by-second machine state timeline (RUN / IDLE / FAULT)
⏱ Automatic downtime & micro-stoppage log with timestamped video proof
📈 Real-time Availability calculation for OEE dashboards
🔧 MTBF (Mean Time Between Failures) & MTTR (Mean Time To Repair) analytics for maintenance teams

/02

Machine / Operator
Cycle Time

Whenever the sequence RUN → IDLE → RUN occurs, the system isolates the IDLE segment to measure how long the operator needed to unload the finished part, load the next blank, and restart the machine. A “target swap time” range is entered once per workstation; Vision MES then scores every cycle and highlights speed-ups, slow downs, and drifts.

Outputs:

⏱ Full cycle-time per part (process + operator)
📊 Operator swap-time distribution & trends over time
🚨 Instant alerts when a cycle exceeds the target window
🏆 Operator performance ranking based on swap-time
📈 Data for line balancing, takt-time analysis, and kaizen projects

/03

Personnel (Labour) Analysis

Vision MES counts every person in a defined camera zone and builds a second by-second presence log. When this zone corresponds to a machine or workstation, the system detects exactly when an operator arrives and leaves. By cross-referencing this with machine downtime events, we extract intervention latency—i.e., how long after a fault the operator showed up.

Outputs:

👥 Staffing vs. schedule heat-map with instant over/under-staff alerts
📊 Labour-hours & cost per shift/line for productivity KPIs
⏱ Time-to-intervene metric for downtime root-cause analysis
🔄 Comparison of operator response patterns across shifts or stations
📈 Line-balancing insights based on presence and activity data

/04

Focused Interaction Tracking

Using AI-based hand detection, Vision MES measures how long an operator’s hands remain in a predefined zone—such as a workstation surface, control panel, or fixed part. Each interaction is logged with exact entry and exit timestamps, giving a precise view of where and how long manual work occurs.

Outputs:

✋ Workstation engagement time per cycle or shift
🎛 Operator-panel usage duration & frequency
🔍 Time spent on critical or high-defect components
✅ SOP compliance validation (e.g. did the operator touch the part?)
📈 Detailed input for cycle-time breakdown and process analysis

/05

Scrap Counting

An overhead camera monitors the scrap bin and detects each time an operator’s hands enter the zone. Each entry is assumed to correspond to one discarded part. This simple but effective proxy lets us estimate real-time scrap count without sensors or operator input. Optimal results come from a top-down camera angle for clear visibility.

Outputs:

♻️ Real-time scrap count per shift, operator, or station
📊 Quality leg of OEE calculated from scrap vs. good part totals
🚨 Scrap trend alerts when sudden spikes occur
🔍 Root cause analysis support when paired with machine state data
📈 Shift-to-shift quality comparison for performance tracking and incentives

/06

In-Line Quality Check Verification

When a part reaches a predefined quantity threshold (e.g., every 10 pieces), Vision MES expects the operator to perform a manual quality check at a dedicated inspection table. Using hand detection over this zone, the system verifies whether the operator placed their hands on the surface and remained there for the required duration—serving as proof that inspection occurred.

Outputs:

✅ Quality-check compliance tracking per operator or shift
📜 Audit trail for mandatory inspections and customer requirements
🏆 Competitive differentiation by offering documented inspection proof
🚨 Early alerts if checks are skipped or rushed
📈 Support for certification, traceability, and quality contracts

/07

Raw Material Refill Alert & Response Time

A vision model monitors raw material bins and detects when they are empty. When a bin hits the empty threshold, Vision MES automatically sends a notification—e.g., “Station 4 raw material depleted!”— to the relevant forklift operator or team. The system then measures how long it takes for the new bin to arrive, quantifying internal logistics response time.

Outputs: