Skip to content

Robot Maintenance

Reliability, failures, and maintenance modeling.

Maintenance Types

Scheduled Maintenance

Regular preventive maintenance:

  • Planned intervals
  • Predictable downtime
  • Prevents failures

Unscheduled Maintenance

Reactive repairs after failures:

  • Random occurrence
  • Unpredictable downtime
  • Emergency response

Failure Model

Mean Time Between Failures (MTBF)

Average operating time before failure:

maintenance:
  mtbf_hours: 500  # Fails every ~500 hours on average

Failure Distribution

Exponential distribution (memoryless):

P(failure in next Δt) = 1 - e^(-Δt/MTBF)

This models random failures where:

  • Past operating time doesn't affect future failure probability
  • Failures can happen at any time
  • Rate is constant

Repair Model

Mean Time To Repair (MTTR)

Average repair duration:

maintenance:
  mttr_hours: 2.0  # 2 hours average repair time

Repair Time Distribution

Often lognormal (right-skewed):

maintenance:
  mttr_hours: 2.0
  mttr_stddev_hours: 0.5

Most repairs are quick, some take longer.

Scheduled Maintenance

Interval-Based

Maintenance at regular intervals:

maintenance:
  scheduled_interval_hours: 100
  scheduled_duration_hours: 1.0

Condition-Based

Maintenance triggered by metrics:

  • Operating hours since last maintenance
  • Task count since last maintenance
  • Degradation indicators

Maintenance Process

Failure Workflow

Robot Operating
       │
       ▼
   [Failure]
       │
       ▼
Travel to Maintenance Station
       │
       ▼
   Join Queue
       │
       ▼
  Receive Repair
       │
       ▼
Return to Service

Scheduled Workflow

Maintenance Due
       │
       ▼
Complete Current Task
       │
       ▼
Travel to Maintenance Station
       │
       ▼
   Join Queue
       │
       ▼
Receive Maintenance
       │
       ▼
Return to Service

Availability

Calculation

Availability = MTBF / (MTBF + MTTR)

Example

  • MTBF = 500 hours
  • MTTR = 2 hours
Availability = 500 / (500 + 2) = 99.6%

Fleet Availability

With N robots:

Expected available = N × individual_availability

Maintenance Station

Configuration

maintenance_stations:
  - id: "M1"
    node: "N50"
    bays: 2
    service_time_s: 3600  # 1 hour scheduled

Queue Behavior

Maintenance Station M1 (2 bays):
  Bay 1: [R5] under repair (45 min remaining)
  Bay 2: [R8] scheduled maintenance (30 min remaining)
  Queue: [R11]

Impact on Operations

Reduced Fleet Capacity

During maintenance:

Available robots = Total - Charging - Maintenance

Task Reassignment

When robot fails mid-task:

  1. Task marked incomplete
  2. Robot goes to maintenance
  3. Task reassigned to another robot

Throughput Impact

Effective throughput = Base throughput × Availability

Configuration Example

maintenance:
  # Failure model
  enabled: true
  mtbf_hours: 500

  # Repair model
  mttr_hours: 2.0
  mttr_stddev_hours: 0.5

  # Scheduled maintenance
  scheduled_maintenance: true
  scheduled_interval_hours: 100
  scheduled_duration_hours: 1.0

  # Station assignment
  station_selection: nearest

Reliability Metrics

Key Indicators

Metric Description
MTBF Mean time between failures
MTTR Mean time to repair
Availability Uptime percentage
Failure rate Failures per hour
Maintenance time ratio Time in maintenance / Total time

Fleet Statistics

Fleet Reliability Summary:
  Total robots: 50
  Failures this period: 12
  MTBF (observed): 480 hours
  MTTR (average): 1.8 hours
  Fleet availability: 99.4%

Best Practices

MTBF Estimation

Use historical data:

MTBF = Total operating hours / Number of failures

Maintenance Station Capacity

Plan for peak demand:

Stations needed ≥ Fleet size × (1 - Availability)

Scheduled Maintenance Timing

  • Schedule during low-demand periods
  • Stagger across fleet
  • Don't overload maintenance stations