" "

Guide: how to deploy an autonomous mobile robot fleet with minimal disruption to operations

AMR Deployment Guide featured image

About MOV.AI 

MOV.AI disrupts Autonomous Mobile Robot development with a Robotics Engine Platform that contains everything needed to quickly build, deploy, and operate AMRs in dynamic environments

The adoption of autonomous mobile robots is growing as a result of the significant efficiency and flexibility benefits they deliver.

To reap those benefits, AMRs need to be integrated into an existing industrial environment. Such a process can take up to several months, depending on the complexity of the project and above all, the robot’s capabilities.

The cost of an AMR deployment project is impacted by:

  • Overall project duration
  • The amount of onsite vs. offsite effort
  • Disruption to ongoing operation
  • Customizations

Naturally, the aim is to shorten overall deployment time and speed up time-to-value. However, it is just as important to minimize onsite work, avoiding disruption to ongoing operations and loss of productivity.

Based on MOV.AI’s experience, an AMR deployment project can be reduced to just three days of onsite work. What’s more, ongoing process adaptations and layout changes are achievable by mostly remote, offline efforts.

This guide is divided into three parts:
Section 1: Robot autonomous capabilities and their impact on deployment efforts
Section 2: Automated tools that improve and speed up AMR deployment
Section 3: Tips for ensuring smooth AMR deployment


Section 1: Robots’ autonomous capabilities and their impact on deployment efforts

50 shades of autonomy and their impact on AMR deployment

Autonomy is more of a category that contains a wide range of abilities rather than a specific, well-defined capability. Let us explain:
While in the past, automated guided vehicles that follow floor markings were considered the height of automation, today most advanced AMRs are expected to navigate using SLAM. But navigation is just one element in the ability to perform a complex task.
To complete a task, an AMR needs to navigate to the right location but also perform many other actions to complete the task:

  • Avoid unexpected obstacles along the way, some of them human
  • Maneuver safely around corners when fully loaded
  • Operate in an environment with patchy internet coverage
  • Find the correct cart or pallet to pick up
  • Identify the correct drop-off location
  • Execute fine procedures and movements to pick up a pallet or a cart
  • Collaborate with humans, IOT devices, and other machines

Warehouses, especially ones with humans operating in them, are highly dynamic if not downright chaotic environments. In addition to the stochastic nature of human behavior, business processes are changing at a growing speed in an attempt to meet changing business environments and needs.

Developing highly autonomous AMRs that adapt quickly and easily and perform fine maneuvers is a complex task. Not many can perform many of the above-mentioned activities without external aids, let alone all of them.

All AMR deployments require some form of physical adaptation of the operational environment to support AMR operations, but some require more effort than others. Depending on the robot’s capabilities, adaptations may include:

  • Charging stations or docking areas for the robots
  • Set up of pick-up and drop-off points or areas, sometimes with specialized positioning infrastructure
  • Navigation aids: floor markings or other navigation aids such as AprilTags, RFID tags, or magnetic strips to guide the AMRs along predefined paths
  • Network enhancements to ensure coverage: AMRs vary in their coverage demands based on their autonomy capabilities.

The more extensive the physical changes to the warehouse environment, the more disruptive they are to ongoing operations, and the more difficult it is going to be to make changes down the line.

There are several tradeoffs to consider in AMR selection, but these are the top two:

Cost vs. flexibility

A robot intended to work in a dynamic environment without infrastructure adaptation or separation from humans, needs to be capable of working in such an environment. Higher accuracy, better navigation capabilities, and advanced decision-making capabilities are required, usually implying a higher cost.

Fixed operation vs. programmability 

Flexible robots may require less upfront investment in infrastructure, but the rules and tasks do need to be defined, configured, and redefined when a business change occurs. This requires either programming capabilities or extensive automated setup and planning tools.

Section 2: Automated deployment tools

Autonomous deployment tools play a critical role not only in speeding up AMR deployments but also in ensuring accuracy and efficiency. Remote deployment tools also ensure there is no interference with operations.

Before getting into the tools, let’s review the AMR deployment process.

AMR Deployment Process

A Diagram of the AMR Deployment Process

Project Planning

The initial planning and high-level definition of the project, including business objectives, high-level task definition, selection of the robots, and high-level definition of the integrations into the automation environment.

Offline Setup

Mapping: Prepare and validate the site map using CAD and automated onsite mapping. This is the only stage in the offline setup that may interfere with the facility’s ongoing operation, as it involves the placement of navigation aids and (AprilTags) and 2-3 hours of mapping, depending on the size of the site.

Network: Define critical areas and validate coverage. Note that robots with distributed computing and decision-making capabilities do not require full coverage is not compulsory, save for specific key points.

Define operational data on the scene: Pickup and drop-off areas, charging areas, waiting zones and no-through zones. Visual scene editing tools create a huge difference in the speed and accuracy of this stage.

Design traffic guidelines: Define the traffic management tools and rules, fixed routes (if used), areas of free navigation, areas of limited or restricted access, speed limits, and safety rules.

Setup tasks and support logic: Define the series of actions that AMRs should take. Mode of operation (on-demand, fully automated), scheduling logic, task allocation logic (proximity, battery, traffic, etc., or automated towards KPI optimization), and triggering source – WMS system or user-triggered

Validation and fine-tuning: This is the critical phase of simulating all tasks and activities off-site using digital twins, to ensure realistic planning and that the offline planning and definition can be realized. It is important to perform this step prior to full on-site deployment, to avoid repeated downtime.
Validations should be run on a functional, process, and business level.

  • Functional level: Run simulations at an AMR level to validate that the AMR performs the defined tasks. Simulation should be as comprehensive as possible, meaning fully simulating the Robot sensors, actuation, and physics.
  • Process level: Run simulations on an AMR and fleet level to validate continuous operation and efficiency. Validate that the pick-up and drop-off points are placed in the correct location and that the AMRs can perform their tasks: that the trajectories and routes make sense, that traffic management rules make sense, and that no traffic jams are created.
  • Business level: Run simulations on a fleet and project level to validate that the plan meets the business and performance KPIs. For example, leverage robot and facility digital twins to run end-to-end simulations and validate that the number of delivered daily units meets the target.

As validation moves on toward the process and business levels, it is less important (nor feasible) to simulate the robot fully; rather, it is more important to simulate the operation, with event/agent-based modeling techniques.

MOV.AI Scene editor
Site planning using a visual scene editor

Onsite Setup

If the previous principles of advanced planning, remote activities, and extensive validation were adhered to, this step should not take long. It should not interfere too much with ongoing operations.

Validate physical preparedness: Route marketing, AprilTags, area markings, and infrastructure adaptation, if needed (ideally, this should even be done before final installation)

Install and configure software: First in a single AMR

Finetune scene: Based on reality
Fleet installation: Install and configure software on the entire fleet

Perform final testing and validation: Mainly of fleet operation

AMR Deployment tools

Here are some deployment tools that help perform these steps accurately and efficiently:

1. Mapping and scene editing tools
    • Automated mapping
    • Visual scene editors
2. Navigation planning tools

Visual tools for trajectory validation, including

    • Identifying sharp turns
    • Trajectories that would result in collisions with static objects
    • Trajectory transitions that require unwanted navigation loops
3. Digital twins and simulation tools to validate deployment setup offsite

Integrated simulation tools and digital twins that allow off-site validation at the functional, process, and business levels.

4. Onsite plan validation and fine-tuning tools

No matter how extensive the planning, there are always gaps between the plan and reality.
For example, the actual position of pick-ups and drop-offs may be different than plans. Trajectories may need to be finetuned. Traffic management rules may need to be adjusted to reality.
Ensure that maps, trajectories, and other rules can be easily validated and adjusted on-site without interfering with operations.

5. Mass installation and configuration tools

To speed up onsite setup and reduce errors

Section 3: Tips for ensuring smooth AMR deployment

We can summarize this section in one short sentence: Plan offline and double-check onsite.

The more preparation work is done before reaching the site, the smoother onsite work will be. However, it is important to keep in mind that there may always be gaps between the plan and reality, be it due to errors or changes that have taken place since initial mapping.

Of course, the robot capabilities and the deployment tools mentioned in the previous two sections help in achieving this.

Our customer success team has come up with the following guidelines that help ensure fast, uneventful deployment:

Avoid obstacles in key areas

Pick-ups and drop-offs are key to having a smooth operation. These locations must be free of any obstacles or clutter inside or in the immediate surroundings.

Network requirements and coverage

Ensure the deployment team has access to the network before going on-site.
Check signal quality along the workflow specifically on the key decision points. Prefer AMR fleets that rely on distributed computing, as these can make decisions even without network connectivity when needed.

Route planning, workflow, and layout definition

Route planning and layout definition is one of the most critical parts when defining a project.
A good layout defines the path your robots can take, notes important areas of interest (like entity pick-up and their respective drop-off points, as well as areas where robot charging stations will be), and sometimes even areas where pedestrian activity is higher.

Here are some guidelines:

Preliminary studies

Gather pictures, videos, and CAD models of the facility so that you all speak the same language. Make sure you define the scope of the project beforehand. Having said that, it is important to map the facility using an AMR to obtain an accurate map as, in our experience, most CAD models are not kept up-to-date

Trajectory planning

Avoid having the same lane for traveling in both directions. Plan either two lanes, one for inbound and another for outbound, or a circular route.

Avoid lanes that have few localization landmarks. Good localization landmarks are fixed objects that never move (e.g., pillars, heavy machinery, elevator shafts, etc.).
Plan waiting areas for your robots if you have a narrow path, e.g., right before the entrance/exit to an elevator, to avoid traffic deadlocks.

Parking and pickup zones

Specifically in cart-tugging or pallet-moving scenarios, depending on the size of the robot, the size of the cart, and the load, you might find yourself in situations where it is very hard to maneuver in and out of a pick-up or drop-off zone. If so, we propose placing the carts or loads on the diagonal instead of perpendicular to the path, to facilitate the robot’s maneuvering while performing these operations. The required spacing between parked carts and loads should also be considered.

If possible, these zones should be low-traffic areas as the robots need time to scan the areas and to maneuver.

Docking stations

Define where to place the robot’s charging stations. Usually, it is a good idea to have them not far from the robot’s usual work path, like near a pick-up zone. If the facility is very large, you might want to define multiple docking areas.

Route planning

Plan routes to avoid congestion and obstacles. Choose routes with lower traffic. If any other type of automated machinery vehicle shares the same driveway or crosses it, on-site priority rules should be assessed.

Once the layout is well defined, you will be able to map the designated areas of the facility and the scene will be much easier to build, as we will see next.

All site infrastructure work such as marking pick-up zones, drop-off zones, and charging stations should be completed even before onsite setup begins.

Area mapping

Perform the mapping with minimal moving objects or humans on site.
Make sure sensors and cameras are clean and not blocked
Maintain distance in narrow corridors so that the FOV is not occluded and all the correct key points are added

Visualize mapping in real-time, if possible, to identify gaps

Ensure you capture the AprilTags used for localization and robot recovery during the mapping process to simplify accurately depicting their location on the robot map.

Remote access

Ensure robust remote access so that support and updates can be provided from afar. Otherwise, modifying updates and configurations may be cumbersome or even impossible due to a slow internet connection.


  • Test edge case – for example, two robots in the same place.
  • Test one robot and then move on to fleet behaviors to identify traffic conflicts/deadlocks and collaboration.

Training for ongoing success:

  • Working with robots is different from working with humans. When introducing an AMR, there is a learning curve. The more people need to adjust to the robots, the longer the process and the more chance for errors. For example, humans may not even notice a 2-3 cm shift in pallet placement. For an AMR, this may be an insurmountable barrier.
  • Train operators and facility staff on how to describe an issue from a process point of view. This will avoid the need for many onsite visits.



AMR deployment is often a time-consuming and costly process, but it doesn’t have to be. 

Given proper off-site deployment preparation, onsite setup can take as little as three days and does not have to involve significant site infrastructure work.

What’s more, the right robot capabilities and deployment tools can ensure ongoing flexibility to adapt to future changes in business needs and ensure optimal business outcomes.

Talk to us to learn about the robot capabilities and deployment tools available in the MOV.AI Robotics Engine Platform™.

Sign up for news update

*By providing my details I agree to the MOV.AI Terms and conditions and Privacy policy. I also agree to receive emails from MOV.AIand I understand that I may opt out of MOV.AI subscriptions at any time.

Ready to build awesome robots?

We use cookies to so that you can enjoy the best experience on our website. If you continue to use this site we will assume that you are happy with it.
Skip to content