Predictability beats improvisation in mobile robot fleets
Guest blog by Nicola Tomatis, CEO of BlueBotics (MHI Member Company)
For years, the materials handling industry has pursued a compelling vision: autonomous vehicles that can navigate anywhere, adapt to any environment, and improvise through complex situations. The appeal is obvious. It promises facilities where fleets of mobile robots, acting more like self-driving cars, dynamically find new routes whenever something blocks their path.
But industrial operations are not open road networks where vehicles can freely explore alternative routes. Operators design these environments for repeatability, timing, and coordination. They need materials to arrive at precise locations within defined cycles, shared spaces to remain predictable, and traffic flows to stay stable enough to keep throughput on track.
That is why industrial navigation is shifting away from unlimited autonomy and toward b. In practice, operators and system designers define virtual paths that optimize movement through the facility while allowing vehicles to execute pragmatic obstacle-avoidance maneuvers when needed.
The Limits of Improvisation
That tension sits at the heart of the long-running debate between traditional Automated Guided Vehicles (AGVs) and Autonomous Mobile Robots (AMRs), often framed as a contest between rigidity and flexibility. Modern AGVs follow predefined virtual paths and typically stop when an obstacle blocks their route, alerting an operator. AMRs, by contrast, promise greater autonomy by detecting obstacles and dynamically selecting alternative routes.
In theory, that flexibility should improve efficiency. In practice, however, it often creates new challenges as fleets scale.
Industrial sites are shared environments where dozens or even hundreds of vehicles operate around people, forklifts, and other equipment. When vehicles independently decide how to navigate around obstacles, they can quickly make traffic patterns unpredictable. They turn intersections into contested zones, create bottlenecks in narrow aisles, and let local navigation decisions disrupt the wider fleet.
The challenge is not obstacle detection. Modern mobile robots can already identify objects in their path. The real challenge is making them coordinate their behavior across an entire fleet operating in constrained spaces. When vehicles act without a common set of rules, local decisions can quickly trigger system-level disruption.
Internal logistics depend on predictability. Operators and system designers structure production lines, warehouse flows, and logistics loops around repeatable movements to optimize speed, efficiency, and coordination. When vehicles behave unpredictably, they introduce variability that can disrupt those carefully designed systems.
Understanding Bounded Autonomy
Bounded autonomy addresses this challenge by combining the strengths of structured path-following navigation with the flexibility needed to handle real-world disruptions. System designers and fleet managers keep vehicles on predefined virtual paths as much as possible, helping optimize movement for speed, acceleration, and coordination across the fleet. These paths function much like rail tracks for trains, allowing the system to calculate precisely how vehicles should move through curves, intersections, and shared zones while maintaining predictable traffic patterns.
When a vehicle encounters a blockage, it does not wait indefinitely for an operator to intervene, nor does it abandon its planned route and start navigating freely through the facility. Instead, it carries out a controlled obstacle-avoidance maneuver within predefined limits, moving around the obstruction only in approved areas and only when fleet traffic rules allow, before rejoining its original path as quickly as possible. A bounded-autonomy model can combine virtual-path navigation with controlled obstacle avoidance under fleet-level traffic management.
This approach directly addresses the realities of industrial environments, where exceptions are routine. Pallets may be temporarily left in an aisle, tools may be placed on the floor during maintenance, and short-lived obstacles may interrupt otherwise clear routes. The goal is not to eliminate these disruptions, but to manage them in a way that protects overall system stability.
Efficiency, Deadlock Prevention, and Configurability
To work in industrial settings, this approach relies on three core principles.
For AGV manufacturers and system designers, this means treating obstacle avoidance not as an isolated vehicle function, but as a behavior shaped by fleet-level traffic logic and site-specific operating rules.
The first principle is efficiency. Virtual paths help system designers optimize vehicle behavior in advance by defining routes, turn curvature, and intersection points. That allows vehicles to travel at the maximum safe speed while maintaining smooth acceleration. When a vehicle needs to avoid an obstacle, it takes the shortest feasible path around it within preconfigured limits before returning quickly to its virtual route. In precision-critical areas such as pick-and-drop locations, the system may restrict flexibility altogether.
The second principle is deadlock prevention. While efficiency depends on allowing vehicles to move smoothly, that movement must also remain coordinated across the fleet. Uncoordinated obstacle avoidance is one of the most common causes of gridlock in autonomous fleets. A more robust model brings avoidance maneuvers under the fleet’s traffic management framework, allowing the fleet manager to prevent conflicting actions in the same space. One key rule is that vehicles may move around objects, but not around other vehicles.
The third principle is configurability. No two industrial environments are identical, so navigation systems must respond to local operating requirements. For AGV manufacturers and system designers, that makes configurability essential. They need to define how far a vehicle may deviate from its virtual path, where avoidance maneuvers are permitted, and how each vehicle should respond when it encounters an obstacle.
System Performance Over Vehicle Intelligence
The practical value of this model becomes clear in a common industrial scenario: a pallet temporarily left in a wide aisle. In that situation, a traditional AGV following a fixed path would stop and wait for the obstruction to be removed. A vehicle using pure AMR-style obstacle avoidance, by contrast, might try to find an alternative route, but that detour could create congestion elsewhere in the fleet.
With a virtual-path-based approach, the vehicle pauses briefly, performs a controlled avoidance maneuver within predefined limits, and then returns to its original route. As a result, the system resolves the obstacle while restoring predictable traffic behavior as quickly as possible.
This points to a broader lesson from mobile robotics deployments: system-level performance matters more than the intelligence of any single vehicle. A robot that can improvise around obstacles may look impressive on its own. But when large fleets operate in the same environment, coordination, traffic management, and predictable behavior matter far more than individual autonomy.
Predictability as an Operational Advantage
Industrial automation exists to deliver materials to the right place at the right time. In environments where production cycles run every minute or less, even minor disruptions can quickly have significant operational consequences.
Predictability, therefore, becomes an operational advantage. For fleet architects and system designers, the advantage is that vehicle behavior can be governed at the system level rather than left to individual machine decisions. For end users and operators, this translates into workflows that are easier to anticipate, throughput that is easier to protect, and faster recovery during unexpected events.
The result is not a compromise between automation philosophies, but a practical synthesis of structure and adaptability. For those designing industrial robot fleets, it offers a more governable model of autonomy. For those operating them, it provides the predictability needed to maintain stable material flow.
