🎓 Lesson 14
D5
ISO/TS 15066 Risk Assessment for Vision-Guided Cobots
ISO/TS 15066 is a technical specification that tells engineers how to assess and reduce risks when humans work safely alongside robots that use cameras to 'see' and react in real time.
🎯 Learning Objectives
- ✓ Explain the biomechanical limits defined in ISO/TS 15066 for upper-limb impact scenarios
- ✓ Apply the time-to-collision (TTC) formula to calculate minimum safe separation distances for vision-guided cobots operating at variable speeds
- ✓ Analyze camera latency, motion control delay, and robot deceleration profiles to validate compliance with Clause 5.3.2 of ISO/TS 15066
- ✓ Design a risk assessment worksheet aligned with Annex A of ISO/TS 15066 for a vision-guided rock sorting cobot station
📖 Why This Matters
In modern mining operations, vision-guided cobots increasingly assist with real-time ore sorting, drill pattern verification, and blast hole inspection—tasks requiring close human–robot proximity. Unlike traditional industrial robots, these systems respond dynamically to visual input, introducing unpredictable motion timing and reaction delays. Without proper risk assessment per ISO/TS 15066, even low-speed cobots can exceed human tissue tolerance during unexpected contact—leading to musculoskeletal injury or regulatory noncompliance. This standard isn’t optional: it’s the technical backbone for safe integration in ISO 13849-1–based safety architectures.
📘 Core Principles
ISO/TS 15066 introduces two complementary risk assessment pathways: (1) the *power and force limiting* approach, grounded in biomechanical data from ISO 13857 and ECRI studies; and (2) the *speed and separation monitoring* (SSM) approach, which requires calculating minimum safe distances based on detection system performance. For vision-guided cobots, the SSM pathway dominates—because camera frame rate, image processing latency, controller cycle time, and robot deceleration capability collectively determine the system’s ability to stop before harmful contact. Crucially, the standard mandates worst-case latency aggregation—not just camera exposure time—and treats vision pipelines as integrated sensing–actuation loops, not isolated components.
📐 Time-to-Collision (TTC) for Vision-Guided SSM
The TTC model determines the minimum distance required between a human and a moving cobot to allow full stop upon detection. It accounts for total system response time—including vision acquisition, processing, decision, and mechanical deceleration. This formula is mandatory for Clause 5.3.2 validation in ISO/TS 15066.
Minimum Safe Separation Distance (S)
S = (v × tₜₒₜₐₗ) + (v² / (2 × a))Calculates the shortest distance ensuring full stop before contact, given constant velocity, total system latency, and uniform deceleration.
Variables:
| Symbol | Name | Unit | Description |
|---|---|---|---|
| S | Minimum safe separation distance | m | Distance between human and cobot at detection initiation |
| v | Maximum approach speed | m/s | Highest speed the cobot may attain toward the human zone |
| tₜₒₜₐₗ | Total system response time | s | Sum of vision acquisition, processing, decision, and command transmission latencies |
| a | Deceleration capability | m/s² | Robot’s verified worst-case stopping deceleration under load |
Typical Ranges:
Conveyor-side vision sorting cobot: 0.25 – 0.65 m
Drill pattern verification drone-mounted cobot: 0.8 – 1.5 m
💡 Worked Example
Problem: A vision-guided cobot sorts rocks on a conveyor at 0.8 m/s. Its stereo camera has 60 fps (16.7 ms frame interval), embedded AI inference latency = 22 ms, PLC scan time = 10 ms, safety controller reaction = 15 ms, and robot deceleration = 1.2 m/s². Calculate minimum safe separation distance (S).
1.
Step 1: Sum all latency components: 16.7 + 22 + 10 + 15 = 63.7 ms = 0.0637 s
2.
Step 2: Compute distance traveled during latency: S₁ = v × t = 0.8 × 0.0637 = 0.051 m
3.
Step 3: Compute stopping distance using v² = u² + 2ad → d = u²/(2a) = (0.8)²/(2×1.2) = 0.267 m
4.
Step 4: Total S = S₁ + S₂ = 0.051 + 0.267 = 0.318 m
Answer:
The minimum safe separation distance is 0.318 m, which exceeds the ISO/TS 15066 recommended lower bound of 0.3 m for light contact on the forearm — confirming compliance for this scenario.
🏗️ Real-World Application
At BHP’s Jimblebar iron ore operation (Pilbara, WA), a vision-guided UR10e cobot verifies blast hole depth and collar position using a calibrated ZED 2i stereo camera before drilling commences. Engineers applied ISO/TS 15066 to quantify cumulative latency across NVIDIA Jetson inference (18 ms), ROS2 safety node (12 ms), and UR e-Series brake response (210 ms worst-case). They calculated S = 0.92 m and installed physical light curtains at 1.1 m—adding 0.18 m safety margin per ISO 13855. Post-deployment HMI logs confirmed zero false positives and <0.3% missed detections during 12,000+ operational cycles.
🔧 Interactive Calculator
🔧 Open Industrial Robot Vision Guidance & Integration Calculator📋 Case Connection
📋 Vision-Guided Palletizing Robot for Mixed-SKU E-Commerce Fulfillment
Unstructured tote input with random part orientation, variable box dimensions, and reflective surfaces causing glare
📋 Vision-Guided Arc Welding Cell for Automotive Chassis Assembly
Thermal distortion-induced seam deviation (>1.8 mm) between stamped aluminum panels; inconsistent joint gap due to fixtu...
📋 3D Vision-Guided Bin-Picking for Aerospace Fastener Kits
Highly reflective titanium fasteners (M3–M8), nested geometry, dense packing, and strict traceability requirements
📋 Vision-Guided Deburring Robot for Cast Aluminum Engine Blocks
Variable burr height (0.1–2.4 mm) and location due to inconsistent casting; complex freeform surfaces limiting fixed-too...