Skip to content

Batching

Grouping tasks for improved efficiency.

What is Batching?

Batching combines multiple tasks into a single unit, allowing a robot to complete several tasks in one trip.

Without Batching

Task 1: Pick SKU-A from R1 → Station S1
Task 2: Pick SKU-B from R2 → Station S1
Task 3: Pick SKU-C from R3 → Station S1

Robot trips: 3 (one per task)

With Batching

Batch: [Task 1, Task 2, Task 3]

Robot trip: R1 → R2 → R3 → S1

Robot trips: 1 (all tasks together)

Why Batch?

Reduced Travel

Without batching:
  Travel = 3 × (rack to station) = 45m

With batching:
  Travel = R1→R2 + R2→R3 + R3→S1 = 25m

Savings: 44%

Higher Throughput

  • Fewer trips = more capacity
  • Less congestion from robot movements
  • Better robot utilization

Trade-off

Batching delays individual tasks:

Without batching:
  Task 1: Complete at t=15

With batching (wait for batch):
  Task 1: Complete at t=30

Individual task delayed, but total work is faster

Batching Strategies

Order-Based Batching

Group tasks from same order:

policies:
  batching:
    type: order_based

Order O1: SKU-A, SKU-B, SKU-C

Batch = [Pick SKU-A, Pick SKU-B, Pick SKU-C]

Pros:

  • Natural grouping
  • Completes orders faster

Cons:

  • Small orders don't benefit
  • May wait for large orders

Zone-Based Batching

Group tasks from same zone:

policies:
  batching:
    type: zone_based
    zones:
      zone_a: [R1, R2, R3]
      zone_b: [R4, R5, R6]

Tasks in Zone A:
  Pick from R1, Pick from R2, Pick from R3

Batch = All Zone A picks

Pros:

  • Minimizes travel within batch
  • Works across orders

Cons:

  • Orders may span zones
  • Requires zone definition

Proximity-Based Batching

Group nearby tasks:

policies:
  batching:
    type: proximity
    max_distance: 10.0

Task 1: R1 (position 5,5)
Task 2: R2 (position 6,7)
Task 3: R5 (position 50,50)

Batch 1 = [Task 1, Task 2] (close)
Batch 2 = [Task 3] (far from others)

Pros:

  • Optimizes travel dynamically
  • No predefined zones needed

Cons:

  • More complex to compute
  • Batch composition varies

Time-Window Batching

Group tasks arriving within time window:

policies:
  batching:
    type: time_window
    window_s: 30.0

t=0:  Task 1 arrives
t=10: Task 2 arrives
t=20: Task 3 arrives
t=30: Window closes

Batch = [Task 1, Task 2, Task 3]

Pros:

  • Simple to implement
  • Predictable batching

Cons:

  • Delays early tasks
  • Fixed window may not match demand

Batch Parameters

Maximum Batch Size

policies:
  batching:
    max_batch_size: 5

Limits items per batch:

Queue: 8 tasks
Max size: 5

Batch 1: 5 tasks
Batch 2: 3 tasks

Maximum Wait Time

policies:
  batching:
    max_wait_time_s: 60.0

Don't wait too long to fill batch:

t=0:  Task 1 arrives (batch started)
t=60: Timer expires (batch released with 2 tasks)

Minimum Batch Size

policies:
  batching:
    min_batch_size: 2
    min_wait_time_s: 10.0

Wait for minimum before releasing.

Batch Execution

Sequence Planning

Optimize visit order within batch:

Batch: [R3, R1, R5]

Nearest neighbor sequence:
  Start → R1 → R3 → R5 → Station

Better than: Start → R3 → R1 → R5 → Station

Partial Completion

Handle failures gracefully:

policies:
  batching:
    partial_completion: true

If one pick fails, continue with rest.

Configuration Example

policies:
  batching:
    enabled: true
    type: proximity

    # Size limits
    max_batch_size: 6
    min_batch_size: 2

    # Time limits
    max_wait_time_s: 45.0
    min_wait_time_s: 10.0

    # Proximity threshold
    max_distance: 15.0

    # Execution
    sequence_optimization: nearest_neighbor
    partial_completion: true

Batching Metrics

Batch Statistics

Metric Description
Avg batch size Mean tasks per batch
Batch fill rate Actual / Max size
Wait time for batching Delay before batch starts

Efficiency Metrics

Metric Description
Travel per task Total travel / Tasks
Throughput Tasks per hour
Travel savings vs. non-batched

When to Batch

Good for Batching

  • High task volume
  • Tasks with common destinations
  • Spatially clustered tasks
  • Non-urgent tasks

Avoid Batching

  • Time-critical tasks
  • Sparse task arrivals
  • Random spatial distribution
  • Very low volume

Testing Batching

Compare With/Without

waremax compare scenario.yaml \
  --param policies.batching.enabled=false \
  --param policies.batching.enabled=true

Tune Parameters

waremax sweep scenario.yaml \
  --param "policies.batching.max_batch_size=[3,5,7,10]" \
  --param "policies.batching.max_wait_time_s=[30,60,90]"

Metrics to Watch

  • Travel per task (lower = better batching)
  • Task latency (watch for excessive delays)
  • Throughput (should improve)