How Warehouse Robots Are Controlled: Three Levels of Control | YaCuAi

How Warehouse Robots Are Controlled: Three Levels of Control

In modern automated warehouses, dozens or even hundreds of robots operate simultaneously. To understand how such a complex system is organized, it is convenient to view the control architecture in three levels: upper, middle, and lower. They are not ranked by importance — each has its own critically important role.

Upper Level: WMS as the Strategic Brain

Warehouse Management System (WMS) is the main planner of all warehouse processes. Imagine that a truck with pallets of goods has just arrived at the warehouse. At this moment, the WMS:

  1. Analyzes the parameters of each item: dimensions, shelf life, storage conditions

  2. Checks product compatibility (for example, chemicals cannot be stored near food)

  3. Evaluates turnover (frequently requested items are placed closer to the shipping area)

  4. Determines optimal storage locations

After that, the system generates a task like: "Move pallet ID-45678 from the receiving area to location C-05-12." The WMS does not care who executes this task — a human with a forklift or an autonomous robot. Its role is optimal planning, not execution control.

Middle Level: RMS — The Conductor of the Robotic Orchestra

Robot Management System (RMS) is the intermediate layer that transforms abstract WMS tasks into specific robot actions. If there are only a few robots, the task is simple. But when there are hundreds, it becomes a real logistical puzzle:

  • How to distribute tasks among robots?

  • How to build collision-free routes?

  • What if multiple robots need to pass through a narrow aisle at the same time?

  • When should robots be sent for charging?

RMS solves these problems in real time using:

  • Routing algorithms (e.g., modified Dijkstra algorithm)

  • Priority systems (urgent orders are handled first)

  • Robot state monitoring (battery level, workload)

Interestingly, RMS often uses principles of swarm intelligence, similar to how ants coordinate actions without centralized control.

Lower Level: Where Algorithms Meet Physics

When a specific robot receives a task, the low-level control system comes into play — what is referred to in Russian terminology as the “Driver software stack”. This is essentially the “cerebellum” of the robot, responsible for:

  1. Precise localization — the robot must always know its position with centimeter-level accuracy. This is achieved using:

    • LiDAR

    • Cameras

    • Odometry (wheel rotation tracking)

    • In some cases — magnetic floor markers

  2. Dynamic path planning — if an obstacle appears (a fallen box or a misplaced cart), the robot must navigate around it without disrupting the overall schedule.

  3. Precise positioning — when approaching a pallet, the acceptable error is often no more than 5 mm.

At this level, PID controllers operate, adjusting movement every millisecond while accounting for:

  • Current position

  • Speed

  • Inertia

  • Surface friction

Why Is It More Complex Than It Seems?

Each of the three levels represents an entire domain:

  • WMS operates with big data and complex optimization algorithms

  • RMS solves NP-hard routing problems in real time

  • The low-level system deals with physical constraints and stochastic disturbances

At the same time, all three levels must work synchronously. A delay of just 100 milliseconds at one level can trigger a cascading failure across the entire system. That is why modern automated warehouses are not just “robots replacing humans,” but complex cyber-physical systems where every detail matters.

Присоединяйся к нам в telegram
  • лайф наших роботов
  • самые крутые пилоты
  • говорим всю правду о разработке
Подписаться
Интерфейс