A microfactory line singulates mixed waste on a conveyor, identifies each item with a near-infrared / RGB vision sensor, then uses a pneumatic pick-and-place arm to eject classified items into material-specific bins. Throughput is limited by whichever stage is slowest — here, the arm's mechanical cycle time.
item_gap = 1.2 m
spawn_dt = item_gap / belt_speed
p_correct = sensor_accuracy
arm_ready = (t - t_last_pick) >= cycle_time
sorted = arm_ready AND classified
purity = correctly_binned / total_binned
throughput = items_binned / elapsed_time
- Belt speed — sets how densely items arrive at the sensor and arm.
- Sensor accuracy — probability the vision system labels a material correctly; misclassified items land in the wrong bin.
- Arm cycle time — minimum time between pick actions; items arriving while the arm is busy fall through to reject.
- Waste stream mix — changes the material ratio entering the line, affecting how often each bin fills and how contamination compounds.
Real microfactories balance exactly this tradeoff: faster belts raise raw throughput but overwhelm a fixed-cycle-time actuator, dropping both effective recovery rate and purity.