Priority¶

Task and robot priority systems.

Why Priority?¶

Not all tasks are equal:

Urgent orders need faster processing
VIP customers expect better service
Some robots may have special capabilities

Priority ensures important work gets done first.

Priority Levels¶

Numeric Priority¶

Higher number = higher priority:

Priority 10: Emergency/VIP
Priority 7:  Express orders
Priority 5:  Standard orders (default)
Priority 3:  Low priority
Priority 1:  Background tasks

Named Levels¶

priorities:
  critical: 10
  high: 7
  normal: 5
  low: 3
  background: 1

Task Priority¶

Assignment¶

Tasks receive priority from:

Order-based:

orders:
  - order_id: "O1"
    priority: 8  # High priority order

Time-based:

policies:
  priority:
    type: due_time
    urgent_threshold_s: 300  # 5 min before due
    urgent_boost: 3

Rule-based:

policies:
  priority:
    rules:
      - condition: "sku.category == 'perishable'"
        priority: 8
      - condition: "customer.tier == 'vip'"
        priority: 7

Dynamic Priority¶

Priority can change over time:

Task created:     Priority 5
30 min remaining: Priority 6
10 min remaining: Priority 8
Overdue:          Priority 10

Priority Effects¶

Task Allocation¶

Higher priority tasks assigned first:

Queue:
  Task A (priority 7)  ← Assigned first
  Task B (priority 5)
  Task C (priority 3)

Station Queue¶

Priority affects queue position:

policies:
  station_queue:
    type: priority_based

Station Queue (with priority):
  Position 1: R1 with priority 8 task
  Position 2: R2 with priority 7 task
  Position 3: R3 with priority 5 task (arrived first)

Traffic Priority¶

Higher priority robots may have right-of-way:

policies:
  traffic:
    priority_routing: true

Priority Schemes¶

FIFO (First In, First Out)¶

No priority consideration:

policies:
  priority:
    scheme: fifo

Queue order = Arrival order

Strict Priority¶

Always process highest priority first:

policies:
  priority:
    scheme: strict

Priority 10 always before Priority 9
Priority 9 always before Priority 8
...

Risk: Low priority tasks may starve.

Weighted Fair¶

Balance priority with fairness:

policies:
  priority:
    scheme: weighted_fair
    weight_factor: 2.0

Higher priority gets more service but doesn't completely block lower priority.

Aging¶

Priority increases with wait time:

policies:
  priority:
    scheme: aging
    age_factor: 0.1  # +0.1 priority per minute

Task at t=0:  Priority 3
Task at t=10: Priority 4 (aged)
Task at t=20: Priority 5 (aged more)

Prevents starvation of low-priority tasks.

Preemption¶

Concept¶

Higher priority task interrupts lower priority:

R1 executing Task A (priority 3)
Task B arrives (priority 9)

Preemption:
  - Task A paused
  - R1 assigned Task B
  - Task A resumed later (or reassigned)

Configuration¶

policies:
  preemption:
    enabled: true
    min_priority_difference: 3

Preemption Points¶

When can preemption occur:

Point	Description
Any time	Can interrupt mid-task
At node	Only at node arrival
At completion	Only between tasks

policies:
  preemption:
    point: at_node

Robot Priority¶

Static Priority¶

Some robots have permanent priority:

robots:
  - id: "R1"
    priority: 7  # VIP robot
  - id: "R2"
    priority: 5  # Standard

Task-Inherited Priority¶

Robot inherits priority from current task:

policies:
  robot_priority:
    type: task_inherited

R1 with priority 3 task → Robot priority 3
R2 with priority 8 task → Robot priority 8

Traffic conflict: R2 wins

Configuration Example¶

policies:
  priority:
    # Base scheme
    scheme: aging
    age_factor: 0.1

    # Task priority sources
    default_priority: 5
    due_time_boost: true
    urgent_threshold_s: 300
    urgent_boost: 3

    # Preemption
    preemption_enabled: true
    min_priority_difference: 4
    preemption_point: at_node

    # Robot priority
    robot_priority: task_inherited

    # Traffic
    traffic_priority: true

Priority Metrics¶

Task Metrics¶

Metric	Description
Avg completion by priority	Time by priority level
Priority violations	High priority delayed
Starvation count	Tasks waiting too long

Fairness Metrics¶

Metric	Description
Wait time variance	Spread of wait times
Completion rate by priority	Tasks completed per level
Preemption count	How often preemption occurs

Best Practices¶

Don't Overuse High Priority¶

If everything is priority 10:
  → Nothing is really high priority
  → System behaves like FIFO

Reserve high priority for truly urgent tasks.

Use Aging for Fairness¶

Prevents low-priority starvation:

policies:
  priority:
    scheme: aging
    age_factor: 0.1
    max_priority: 9  # Cap priority growth

Monitor Priority Distribution¶

Track how priorities are distributed:

Priority Distribution:
  10: ██ (5%)
  7-9: ████ (15%)
  4-6: ████████████████ (60%)
  1-3: █████ (20%)