Unplanned robot downtime costs manufacturers an average of $260,000 per hour, according to Aberdeen Group research. In warehousing, a single AMR going down during peak shifts can cascade into hundreds of delayed orders. Yet most organizations treat robot maintenance as an afterthought — reacting to failures instead of preventing them.
This guide covers how to build a maintenance program that keeps robots running and operations profitable.
Preventive vs. Reactive vs. Predictive Maintenance
Three maintenance strategies exist. Most successful operations use a combination of all three.
Reactive maintenance (break-fix): Fix it when it breaks. This is the default for organizations without a maintenance program. It's the most expensive approach — repair costs are 3-5x higher than planned maintenance because failures cascade (a worn bearing destroys a gearbox), and downtime is unscheduled and disruptive.
Preventive maintenance (PM): Scheduled inspections and part replacements at fixed intervals — regardless of whether the component shows wear. This is the foundation of any maintenance program. PM programs reduce unplanned downtime by 25-35% compared to reactive-only approaches.
Predictive maintenance (PdM): Using sensor data, vibration analysis, thermal imaging, or AI models to predict failures before they occur. PdM optimizes PM by replacing parts based on condition rather than schedule — reducing unnecessary replacements while catching failures earlier. PdM adds 10-20% improvement on top of PM alone, but requires data infrastructure and expertise.
For most operations, start with PM and layer PdM as your data and capabilities mature.
Designing a Preventive Maintenance Schedule
Build your PM schedule from three sources: vendor recommendations, manufacturer data, and your own operating experience.
Daily checks (operator-level, 5-10 minutes per robot):
- Visual inspection for physical damage, loose connections, debris
- Check indicator lights and error codes on the HMI
- Verify charging behavior (AMRs) or air supply (pneumatic systems)
- Listen for unusual sounds (grinding, clicking, whining)
- Clean sensors and cameras (dusty or dirty sensors cause navigation errors and safety stops)
Weekly checks (technician-level, 30-60 minutes per robot):
- Inspect cables and harnesses for wear, pinch points, and fraying
- Check end-effector condition (gripper wear, tool calibration)
- Review software logs for recurring warnings or errors
- Test safety systems (e-stops, light curtains, force limiting)
- Verify network connectivity and firmware versions
Monthly checks (technician-level, 2-4 hours per robot):
- Mechanical inspection: joint play, belt tension, chain wear
- Lubrication per manufacturer schedule
- Battery health check (capacity, charge cycles, cell balance)
- Calibration verification (encoder positions, force/torque sensors)
- Backup configuration and program files
Annual checks (technician or vendor, 4-8 hours per robot):
- Complete mechanical overhaul per manufacturer specifications
- Replace wear components on schedule (belts, bearings, seals)
- Full calibration procedure
- Safety system certification
- Software update to current supported version
- Performance benchmark against original specifications
Document everything. A maintenance log that tracks actions taken, parts replaced, and hours logged becomes your most valuable asset for optimizing the schedule over time.
Spare Parts Strategy
Parts availability determines whether a failure costs one hour or one week.
Critical spares (stock on-site): Components whose failure stops the robot completely and have lead times over 24 hours. For a typical cobot: controller boards, teach pendant, motor assemblies, encoder cables. For AMRs: LiDAR sensors, drive motors, batteries, safety scanners. Budget $2,000-$10,000 per robot for critical spares inventory.
Consumable spares (stock on-site): Items replaced regularly through PM. Gripper fingers, vacuum cups, belts, filters, fuses, cable assemblies. Budget $500-$2,000 per robot per year.
Non-critical spares (vendor depot): Components that can be replaced during planned maintenance windows. Cosmetic panels, non-essential sensors, lighting. Rely on vendor or distributor depot stock with 2-5 day delivery.
Parts management tips:
- Negotiate spare parts pricing at time of purchase — vendors offer 10-20% discounts on initial parts kits
- Track consumption rates and adjust inventory quarterly
- Establish min/max levels for each part with automatic reorder triggers
- If running multi-vendor fleets, identify shared components to consolidate inventory
- Keep parts organized and labeled — a $200 part that takes 2 hours to find costs more than the part itself
In-House vs. Contract Maintenance
The right approach depends on fleet size, technical capability, and criticality.
In-house maintenance makes sense when:
- You have 10+ robots justifying a dedicated technician
- Your existing maintenance team has mechatronics skills (or can be trained)
- Response time is critical (24/7 operations, high-cost downtime)
- You want to build institutional knowledge about your specific deployment
Contract maintenance makes sense when:
- You have fewer than 10 robots
- The robot technology is highly specialized (surgical robots, complex vision systems)
- Your maintenance team is already at capacity
- The vendor offers comprehensive service contracts at reasonable rates
Hybrid approach (most common): In-house team handles daily/weekly PM and Level 1 troubleshooting (restart, recalibrate, swap known-good parts). Vendor contract covers Level 2-3 repairs (controller replacement, firmware recovery, mechanical rebuilds) and annual overhauls.
Training investment for in-house maintenance:
- Basic operator maintenance: 8-16 hours per operator
- Level 1 technician: 40-80 hours (vendor training program)
- Level 2 technician: 80-160 hours (advanced vendor training + on-the-job experience)
- Budget $5,000-$15,000 per technician for training courses, travel, and certification
KPIs to Track
You can't improve what you don't measure. Track these metrics monthly.
Overall Equipment Effectiveness (OEE): Availability × Performance × Quality. Target: 85%+ for mature deployments. Most new deployments start at 60-70%.
Mean Time Between Failures (MTBF): Average operating hours between unplanned stops. Higher is better. Track by robot model, by individual unit, and by failure type. If one unit's MTBF is significantly lower than fleet average, investigate.
Mean Time To Repair (MTTR): Average time from failure detection to full operation. Target: under 2 hours for Level 1 issues, under 8 hours for Level 2. Track separately for in-house vs. vendor repairs.
Planned vs. Unplanned Downtime Ratio: Target 80%+ planned (PM) vs. less than 20% unplanned. If unplanned exceeds 30%, your PM program needs adjustment.
Maintenance Cost per Operating Hour: Total maintenance spend / total operating hours. Track over time — it should decrease as your PM program matures and stabilize after 12-18 months.
Parts Inventory Turns: Annual parts consumption / average inventory value. Low turns mean you're overstocked. High turns with stockouts mean you're understocked. Target 2-4 turns per year for robot spare parts.
Read our downtime reduction guide for tactical approaches to improving these metrics, or our maintenance cost reduction guide for strategies to lower spending without compromising reliability.
Frequently Asked Questions
How much should I budget for annual robot maintenance?
Budget 5-12% of hardware cost per year for maintenance. A $50,000 cobot costs $2,500-$6,000 annually to maintain. A fleet of 20 AMRs at $80,000 each should budget $80,000-$192,000 per year. This includes parts, labor, vendor service contracts, and consumables. First-year costs tend to be lower (warranty coverage), while years 3-5 may be higher as components age.
Can my existing maintenance team handle robot maintenance?
If your team has experience with PLCs, servo motors, and industrial networking, the transition is manageable with 40-80 hours of vendor-specific training per technician. If your team is primarily mechanical (HVAC, plumbing, basic electrical), you'll need either a dedicated robotics technician hire or a vendor service contract for anything beyond daily checks.
How do I justify the cost of predictive maintenance technology?
Start with a single high-cost failure mode. If bearing failures on your robot fleet cause $50,000 in annual unplanned downtime, and a $5,000 vibration monitoring system could prevent 70% of those failures, the payback is under 3 months. Scale PdM to the failure modes that cost the most, not across everything at once.
What's the most common cause of robot downtime?
For cobots: collision events and end-effector wear (together about 40% of incidents). For AMRs: navigation sensor issues, network connectivity, and battery degradation (about 50% of incidents). For all robot types: software errors and misconfiguration account for 20-30% of downtime — many of which can be resolved by restart or parameter reset.
How often should I update robot software/firmware?
Apply security patches within 30 days of release. Feature updates should be tested in a staging environment before deployment to production robots. Avoid updating during peak operating periods. Most vendors release 2-4 major updates per year. Skip updates that don't address your specific issues — every update carries a small risk of regression.