Introducing Conveyor Connect: How Chef Robots Communicate With Food Conveyors

Walk into any food manufacturing facility, and you’ll see a mix of automation equipment from different vendors: tray sealers, de-nesters, weighing systems, and conveyors. Each piece of equipment does a specific job, but getting them to work together can be challenging.

‍

Chef robots, for example, need to adapt to whichever conveyor system a food manufacturer is already running on its production lines. Not all conveyors are the same, and we want our robots to be able to deposit on any of them quickly and reliably. 

‍

Additionally, while we want Chef robots to be able to deposit onto every conveyor, one of the key requirements is that the robots can slide onto the line quickly without requiring a large integration (as is common in traditional automation). 

Here’s how we approach each type of conveyor system and ensure our robots communicate quickly with each of them.

‍

The conveyor landscape

Food manufacturing facilities run a wide variety of conveyor systems. The three most common types of meal assembly conveyors are continuous belt conveyors, indexing (stop-and-go) conveyors, and chain conveyors.

• Continuous belt conveyors move trays forward continuously without stopping. Chef robots use an AI system to infer the speed of the continuous belt, identify each tray, track tray positions and orientation in real time, and precisely deposit ingredients.
• Chain conveyors use interlocked chain links instead of a continuous belt—durable, easy to clean, and well-suited for heavier tray loads. For chain conveyors, a direct integration with Chef robots helps ensure more reliable tray tracking and precise deposits.‍
• Indexing (stop-and-go) conveyors use either a belt or a chain, but what sets them apart is how they move. They shift forward, stop, wait for Chef robots to fill trays, then move again. The stop windows can be quite short, so relying solely on visual perception to determine when the conveyor has halted usually slows operations. To solve this, we built a direct line of communication between Chef robots and indexing conveyors.

In any of these cases, the conveyor types can be single or multi-lane. Chef’s AI system can differentiate between lanes and deposit food into one or several of them. 

‍

How Chef robots connect with different conveyors

Getting Chef robots to work reliably on a production line means solving two problems: ensuring the robot knows exactly where each tray is and when to deposit, and, in the case of indexing conveyors, ensuring the conveyor knows when the robot is done before it moves ahead.

‍

For most conveyors—particularly continuous belt systems—Chef relies primarily on AI-driven perception. Our vision system handles tray detection, insert detection, tray tracking, speed estimation, anomaly detection, and overall conveyor understanding (including identifying the conveyor’s 3D plane, its distance from the robot, and any skew or tilt). This perception-first approach enables us to handle the vast majority of real-world edge cases, including slanted or skewed conveyors, non-monotonic velocity changes, routine starts and stops, and variability introduced by upstream or downstream human activity.

‍

But for the other two conveyor types, and to improve Chef’s performance in the tray-tracking problem, we built a robot-to-robot communication system: a wireless radio network that allows Chef robots to share a real-time feed of tray positions and orientations with each other. Once we had the technology for robots to talk to each other, we decided to use a similar wireless radio to communicate with the conveyor. It turns out there are a few meaningful benefits to doing so:

‍

1. Direct control over any conveyor, enabling our robots to:

• Read the conveyor speed in real time and avoid missing trays. Instead of relying solely on camera perception to estimate speed, the robots receive the conveyor’s actual speed from the conveyor controller, making tray position tracking significantly more accurate. This allows Chef robots to: 
  
  • Be able to react better to sudden stops in the conveyor upstream or downstream. 
  • Be able to deposit ingredients more precisely into small inserts, as are common in salads and snack boxes. 
• Set conveyor speed, rather than being locked to whatever was programmed at installation.
• Dynamically adjust conveyor speed to maximize throughput while minimizing missed trays. In other words, if the robot is picking from the back or bottom of the pan, the conveyor can slow down a bit, then speed up a bit when the robot is picking from the front.    

2. Stop-and-go coordination between the conveyor and Chef robots, allowing them to deposit onto indexed conveyors. When the conveyor stops, it signals the companion box, which then tells Chef robots to start depositing ingredients. Once the robots are done, the companion box sends a “done” signal back to the conveyor to ensure the line continues moving.

To make this work, we built the conveyor companion box that 1) was quick to install and honored the Chef’s vision of not requiring any changes to existing line infrastructure, 2) was small, 3) wireless, and 4) allowed Chef robots to still be flexible and move around from station to station and from line to line. The result is a small, waterproof electronics enclosure that attaches to the conveyor’s VFD control system. The companion box contains a radio to enable wireless communication between the conveyor and Chef robots. It can easily integrate with any conveyor and provides very quick integration. The conveyor companion box approach is also scalable; for example, in a facility with multiple conveyors (and thus multiple companion boxes), the user can assign each Chef robot to the correct conveyor directly from the HMI. 

‍

Chef’s perception system plays an important role throughout this entire process. Even after receiving a radio signal indicating that an indexing conveyor has stopped, the robot uses computer vision to verify that a tray is present before depositing any ingredients. It also tracks tray positions and orientations in real time to ensure precise placement.

‍

Real-world deployment at Cafe Spice

One of our customers, Cafe Spice, is already using the conveyor integration feature in production. Initially, when we planned to integrate Chef robots with Cafe Spice’s production lines, the conveyor stop time was quite short for our robots to detect trays and deposit ingredients. The conveyor integration is what made this short window workable. 

‍

If you’re looking to automate your meal assembly lines and want to understand how Chef robots can work with your existing conveyor setup, get in touch with our team.