AGD Intelligence

Portioning and plating of deformable cooked food in high-volume central kitchens

In Compass's central production units (e.g. the Bangalore facility producing 28,000+ meals/day, serving 20 tonnes of cooked food including ~3,500 kg rice, ~3,500 kg lentils and 15,000 idlis daily), prepared food must be portioned and plated/packed into trays and containers for downstream distribution. The task involves grasping, scooping, and depositing soft, sticky, fragile, variable-geometry items — steamed rice, lentils, idlis, sauced components — without crushing, smearing, or under/over-portioning. It sits downstream of bulk cooking and upstream of sealing/transport, on a high-throughput line. It is hard for a robot because each portion deforms unpredictably, items stick and clump, weight-accurate portioning requires real-time force/compliance feedback, and food appearance (plating integrity) is a quality requirement vision alone cannot guarantee. Throughput is very high and consistency is contractually important. We identified this through our own research; we have not confirmed the specifics with the customer directly. This page is our researched read — a starting point for that conversation.

Readiness
stretch
Demand
weak
Source
researched
Failure tol.
medium
Tactile value
high
i

What the task is

RESEARCHED · our reconstruction

In Compass's central production units (e.g. the Bangalore facility producing 28,000+ meals/day, serving 20 tonnes of cooked food including ~3,500 kg rice, ~3,500 kg lentils and 15,000 idlis daily), prepared food must be portioned and plated/packed into trays and containers for downstream distribution. The task involves grasping, scooping, and depositing soft, sticky, fragile, variable-geometry items — steamed rice, lentils, idlis, sauced components — without crushing, smearing, or under/over-portioning. It sits downstream of bulk cooking and upstream of sealing/transport, on a high-throughput line. It is hard for a robot because each portion deforms unpredictably, items stick and clump, weight-accurate portioning requires real-time force/compliance feedback, and food appearance (plating integrity) is a quality requirement vision alone cannot guarantee. Throughput is very high and consistency is contractually important.

To confirm with the customer

Is this the actual task and sequence? What are the real tolerances, cycle rate, and reject criteria, and which steps are today's manual bottleneck? Answering these is what turns this from a researched signal into a validated use case.