Overview
Deploying a robot from lab bench to production site is a 4-8 week process that most teams underestimate. The lab environment is controlled -- consistent lighting, stable network, known objects. The production site has none of these guarantees. This checklist covers every item you need to verify before, during, and after deployment, organized into 8 phases. Print it, assign owners to each item, and track completion.
Phase 1: Site Survey (Week 1)
- 1. Floor plan with robot placement marked (CAD or sketch with dimensions)
- 2. Floor load capacity verified (robot base + payload + table: typically 200-500 kg/sqm)
- 3. Ceiling height verified (minimum: robot reach + 0.5m clearance)
- 4. Ambient temperature range documented (most robots: 5-40C operating)
- 5. Lighting conditions measured (lux meter readings at workspace: target 500-1000 lux)
- 6. Vibration assessment (nearby machinery can affect precision; measure with accelerometer)
- 7. Access paths for equipment delivery (doorway width, elevator capacity)
Phase 2: Safety Assessment (Week 1-2)
- 8. Risk assessment completed per ISO 10218-2 / ANSI/RIA 15.06
- 9. Collaborative workspace boundaries defined and marked on floor
- 10. Emergency stop locations planned (one per access point, within arm's reach)
- 11. Safety fencing/barriers specified (if industrial arm) or PFL limits verified (if cobot)
- 12. Safety signage prepared (ISO 7010 symbols: W012 for industrial robots)
- 13. First aid kit verified at deployment site
- 14. Insurance coverage confirmed for robotic equipment and third-party liability
Phase 3: Infrastructure (Week 2-3)
- 15. Electrical: dedicated circuit verified (20A for most cobots; 30A for industrial arms)
- 16. Electrical: UPS/surge protection installed (recommended for all robots over $5K)
- 17. Network: Ethernet drop at robot location (Cat6 minimum; Cat6a for GigE cameras)
- 18. Network: Dedicated VLAN for robot traffic (isolate from office network)
- 19. Network: Firewall rules configured (allow ROS2 DDS traffic, block external access)
- 20. Network: NTP server configured for time synchronization (<5ms accuracy on LAN)
- 21. Compressed air supply (if pneumatic gripper: 6 bar, 50 L/min minimum)
- 22. Mounting surface prepared (rigid table or floor mount; verify flatness <0.5mm over workspace)
Phase 4: Software Stack (Week 2-4)
- 23. OS installed and hardened (Ubuntu 22.04 for ROS2 Humble; disable unnecessary services)
- 24. ROS2 workspace built and tested on deployment hardware
- 25. Robot driver verified on deployment hardware (not just dev laptop)
- 26. Camera drivers installed and calibrated (intrinsic + extrinsic calibration verified)
- 27. MoveIt2 configuration validated (collision objects match real workspace)
- 28. Policy/application software tested in simulation first
- 29. Logging and telemetry configured (joint states, camera feeds, error logs)
- 30. Auto-start scripts configured (systemd services for ROS2 nodes)
- 31. Remote access configured (SSH, VPN; no public-facing ports)
Phase 5: Integration Testing (Week 3-5)
- 32. Dry run: execute full task cycle without objects (verify motion paths)
- 33. Object testing: verify grasp success on all target objects (minimum 50 trials per object type)
- 34. Edge case testing: unusual object positions, missing objects, double objects
- 35. Recovery testing: trigger each error state and verify graceful recovery
- 36. Emergency stop testing: verify all e-stops function correctly
- 37. Power loss testing: kill power and verify safe restart
- 38. Network loss testing: disconnect Ethernet and verify safe stop
- 39. Endurance test: run continuously for 8+ hours, log any drift or failures
Phase 6: Operator Training (Week 4-6)
- 40. Safety training completed for all personnel with workspace access
- 41. Operator training: start/stop procedure, basic troubleshooting, e-stop recovery
- 42. Maintenance training: daily inspection checklist, cable check, calibration verification
- 43. Training documentation provided (printed quick-start card at workstation)
- 44. Operator certification: each operator demonstrates safe start, operation, and e-stop (documented)
Phase 7: Monitoring Setup (Week 5-7)
- 45. Dashboard configured: cycle time, success rate, error count, uptime (Grafana or SVRC Platform)
- 46. Alerting configured: email/Slack notification on error rate spike or robot stop
- 47. Log rotation and storage: prevent disk full (logrotate for ROS2 logs)
- 48. Backup schedule: configuration files, calibration data, trained models
Phase 8: Go-Live and Incident Response (Week 6-8)
- 49. Incident response plan documented: who to call, escalation path, spare parts location
- 50. Post-deployment review scheduled (2 weeks after go-live): review logs, adjust parameters, collect operator feedback
Typical Timeline
For a standard cobot deployment (UR5e or OpenArm with cameras and gripper), expect 4-6 weeks from kick-off to go-live. For industrial arms requiring safety fencing and PLC integration, expect 8-12 weeks. The critical path is usually site infrastructure (electrical, network) and integration testing.