Can Safety Radar Replace Laser Scanners on Outdoor AMRs?

The Robot Keeps Stopping at the Doorway. Is Radar the Answer?

Consider an illustrative purchasing problem: an autonomous mobile robot travels between a factory hall and a covered loading area. It completes indoor trips reliably, but interruptions increase near the doorway during dusty operations or changing weather. The maintenance team cleans the scanner more often. Operations asks whether a radar upgrade would keep the route moving.

This is where the safety radar vs laser scanner comparison becomes commercially useful. The decision concerns a specific lost capability, a measurable interruption cost and a machine that already has an established control architecture. It cannot be settled by comparing the maximum ranges printed on two product pages.

First establish why the vehicle stops. A dirty optical cover, an airborne particle return, a localization failure, an actual obstacle and a fleet permission timeout can produce similar symptoms for the operator. A new sensor helps only if it addresses the cause. It is also essential to distinguish an unnecessary interruption from a correct protective stop that prevents exposure to a hazard.

The following analysis uses that doorway problem to examine four purchasing choices: improve the existing scanner installation, upgrade the optical solution, add complementary radar, or redesign a protective function around suitable radar equipment. The examples and financial assumptions are illustrative; they are not field results from a particular robot or supplier.

What Actually Changed in the Radar Market During 2026?

Several manufacturer announcements make this a timely procurement question. They concern different functions, however, and should not be treated as interchangeable evidence of readiness.

Safety protection: SICK's April 2026 innovations publication, dated April 21, identifies an extension of its safeRS3 portfolio. It describes configurable three-dimensional protective fields for stationary and mobile hazardous-area protection, with PL d and SIL 2 stated for the system. The publication also identifies resistance to dust, dirt, sparks and ambient light. This is evidence of a commercially developed safety-radar offering, not evidence that every installation or AMR application meets those performance claims. [1]

Configuration: On May 22, Inxpect announced S203A-WT and S203A-WLT sensors with a Teach-In function for field reconfiguration without its configuration software. This development addresses commissioning and changing surroundings. Buyers still need to understand what the function changes and how the resulting configuration is checked before operation. [2]

Navigation perception: Aptiv's June 18 article on AMR perception describes its PULSE sensor combining radar with a wide-angle RGB camera for near-field depth, motion and visual context. The commercial direction is complementary sensing. The article does not, by itself, establish that this perception product can replace a particular personnel-protection system. [3]

Traffic awareness: Pilz announced PSENradar live on September 24, describing presence detection, counting and tracking of moving people and objects in defined zones. This is a recent signal of radar's expanding industrial role. The announcement does not supply the PL, SIL or application-specific evidence needed to approve it as a replacement for an AMR's safety scanner. [4]

The purchasing implication is that the word radar now appears in several different proposals: a protective device, a perception module, a traffic-monitoring product or a configurable sensing subsystem. Ask what decision the quoted product is permitted to authorize before evaluating whether its demonstration looks impressive.

The Replacement Test Starts with a Function Inventory

Before requesting safety radar for AMRs, list every task performed by the existing scanner. One physical housing may deliver a safety-related field status and separate measurement data for navigation. Removing the housing can therefore affect more than the protective stop that originally attracted attention.

Start with the safety function: which hazardous motion must be prevented or stopped, under which operating modes, and using which input and output path? Then list non-safety uses such as localization, obstacle-map updates, docking assistance and diagnostic visualization. Keep those responsibilities visible in the request for quotation.

There is a particularly important standards boundary. The published scope of IEC TS 61496-5:2023 concerns radar-based protective equipment for whole-body detection of adult persons in industrial manufacturing. It explicitly states that aspects of a moving radar protective device application are not considered and that additional consideration may be necessary when a supplier specifies moving use. Therefore, a reference to this specification alone does not resolve AMR suitability. [5]

Request the supplier's documented basis for the proposed mobile application, including permitted mounting, target conditions, motion limits, response behavior and integration requirements. This is a product-and-application question, not an invitation to assume that a component label certifies the entire vehicle. Our guide to safety-rated sensing and navigation perception explains the broader distinction.

A supplier who can fill this inventory precisely is easier to compare than one offering only a general promise of improved robustness. The inventory also reveals whether the proposed change is a small addition or a substantial redesign.

Ask What Happens After the Target Stops Moving

Motion detection, presence sensing and obstacle detection differ

Can radar detect stationary objects? Some radar systems can, but the answer for a purchased system depends on its processing, configuration and intended function. A demonstration of a walking person does not answer whether the same output remains asserted for a person standing still or for a stationary plastic container.

Leuze's LBK product page illustrates why the distinction matters: it describes personnel access and presence monitoring while allowing static objects in the protective field without switching off the safety outputs. That behavior can be useful for the intended application. It also shows why a buyer cannot assume that every safety-radar product acts as a general stationary-obstacle detector. [6]

Software provides another example. Texas Instruments' motion-and-presence demonstration documentation describes a processing stage that removes static clutter by dropping the zero-Doppler bin. This is an implementation example, not a description of all radar products. Its procurement lesson is narrow but valuable: inspect the delivered processing mode rather than inferring behavior from the radio hardware alone. [7]

Separate three requirements in the quotation. One concerns an approaching person. Another concerns a person who has entered and remains within the relevant area. The third concerns a motionless material obstacle. Write down the required response for each instead of using the ambiguous phrase detects everything.

A stationary obstacle looks different from a moving sensor

When a radar moves with the vehicle, a stationary object can have relative motion with respect to the sensor. The observed motion also changes with the direction of approach. A successful test with a radar fixed to a bench therefore does not establish identical behavior while the AMR turns, reverses, stops or begins moving again.

For the doorway example, ask the supplier to explain a complete sequence: the AMR approaches a stationary object, stops before it, waits, and requests permission to move again. Which function keeps the stop active? Does the output depend on recent motion? What information survives a controller restart? A clear answer should connect documented behavior to the actual sequence.

Keep response time separate from restart timing

Inxpect's S203A-WT page provides a useful example of terminology. It lists a four-second minimum restart timeout and describes static-object detection as an option in the area where restart prevention is active. Those details are specific to the product and function. The restart timeout must not be copied into a stopping-distance calculation as though it were the sensor's detection response time. [8]

Request separate values and definitions for detection response, output update, communication delay, fault reaction and restart conditions. If the supplier supplies only one timing number, the comparison is incomplete.

Dust Tolerance Is Valuable, but It Is Only One Buying Criterion

Teams investigating radar sensors for dusty environments are usually trying to preserve useful operation when an optical sensing path becomes unreliable. That is a sensible investigation. The required outcome remains reliable detection of the relevant targets, not simply continued sensor operation.

A specification should distinguish airborne dust from deposits on the sensor enclosure, occasional mist from sustained wetting, and a brief transition from a full outdoor shift. State how each condition will be characterized during evaluation. Otherwise, both suppliers can claim a successful dust test while having tested materially different conditions.

Target visibility depends on more than physical size

Inxpect's technical explanation of radar cross section identifies material, size, wavelength, polarization and viewing geometry as factors affecting the energy returned to a radar. A target that appears large to a person is therefore not automatically an easy radar target in every orientation. [9]

For procurement, turn that principle into a representative target list. Include actual empty and loaded carriers, the relevant packaging, and material obstacles that could obstruct the approved route. Record orientation and placement. A large metal demonstration reflector can help check a setup, but it is not a substitute for the objects the AMR must handle in production.

Personnel-protection evidence also needs to remain separate from material-detection evidence. A test showing that a box is detected says little about the defined human-detection function. Conversely, evidence for adult whole-body detection does not establish detection of every thin protrusion or small item on the floor.

Range accuracy does not tell you whether two targets remain separate

Texas Instruments' application brief distinguishes range accuracy from range resolution, and angular accuracy from angular resolution. Accuracy concerns the estimated position of a target; resolution concerns separating nearby targets. That distinction is essential when comparing radar vs lidar for mobile robots. [10]

Imagine a carrier beside a rack upright at the loading-area entrance. The buying question is whether the supplied system can represent the occupied space usefully enough for its assigned task. A precise distance estimate to one strong return does not answer whether the neighboring object remains distinguishable. Ask for the relevant scene, not only an isolated-target accuracy chart.

Similarly, a broad field of view can increase coverage while creating a need to distinguish the travel corridor from nearby activity. Compare the usable field shape, boundary behavior and mounting restrictions. A three-dimensional field is valuable only when its approved geometry matches the hazard and does not create unacceptable operational interference.

Evaluate the installed assembly

The quotation should cover the sensor as mounted on the robot, with its brackets, covers, cable routing and normal payloads. Ask the supplier whether any proposed cover is approved, whether it changes performance, and which installation checks are required. Do not transfer a bare-sensor demonstration result to a different enclosure without supporting evidence.

Keep difficult reflective surroundings, nearby machinery and other active sensors in the evaluation scene. The purpose is to establish the limits of this installed system. It is not to award either technology a universal all-weather label.

Walk the Proposed Upgrade Through One Complete Route

A route-based discussion often exposes missing requirements faster than a long list of sensor features. For the illustrative factory-to-loading-area journey, examine five locations with the supplier. Treat these as procurement scenarios rather than a complete safety-validation procedure.

Inside the hall

The existing AMR may use scanner measurements to recognize its surroundings. If radar replaces that scanner's protective function, ask whether the original scanner remains necessary for localization. A proposal that retains it may still deliver useful protection or availability, but the hardware-removal savings disappear. Price the architecture that will actually be installed.

At the doorway

Observe the transition itself, including changing background, door movement and any approved change in operating mode. Establish who authorizes the new mode and how the controller knows it is active. A sensor capable of working on both sides of a doorway does not automatically provide a correct transition between two different protective configurations.

Along the covered outdoor lane

Compare the relevant weather conditions, surface conditions, traffic and payload configurations. If the robot must slow down to remain inside its validated operating limits, include that speed in the throughput comparison. A route that no longer stops unnecessarily may still deliver little additional capacity if the new operating speed is substantially lower.

Separate perception limits from traction limits. An optical upgrade or radar may improve detection while a wet or contaminated surface still changes the vehicle's ability to stop. The approved route must remain within both sensing and braking limits. A successful low-visibility demonstration is not evidence that speed can remain unchanged on every surface.

Near waiting personnel and parked carriers

Request evidence for the specified detection function after motion ceases. A system that tracks moving traffic may solve congestion analysis, while a protective system has a different responsibility. Make sure the proposal explains how it treats waiting, obstruction and temporary disappearance from view without translating uncertainty into permission to move.

At the destination and on the return trip

Include the approach to the station, the normal stop, any load transfer and the reverse or turning motion needed to leave. The configuration used during straight travel may not cover the same exposure during departure. A successful outbound trip is therefore only part of the evidence.

Use the site's existing environmental edge-case testing process for detailed challenge conditions and records. The additional purpose here is to identify which responsibilities move from one sensor or subsystem to another.

Safety Radar vs Laser Scanner: Four Upgrade Paths to Compare

Path one: correct the diagnosed problem

Check whether the current sensor is being used within its documented limits and whether the installation matches the approved design. A supplier-supported cleaning arrangement, corrected mounting or repaired diagnostic issue may be the most effective investment. Establish the baseline again after the correction, so the radar proposal is compared with a properly functioning system.

This path is particularly relevant when interruptions began after a payload change, mechanical impact or maintenance intervention. Avoid turning a repairable defect into an expensive technology migration.

Path two: evaluate an appropriate optical upgrade

Outdoor operation does not automatically eliminate laser scanners. In its September 2024 technical interview, SICK describes outdoorScan3, an airWiper accessory and a Dynamic Weather Assist system combining the scanner, controller and software. These are examples of optical-system approaches to challenging conditions; their applicable limits still need verification. [11]

For buyers comparing outdoor AMR safety sensors, this creates a meaningful alternative quotation. Ask whether an optical upgrade preserves existing measurement interfaces and reduces integration effort. Our safety laser scanner selection guide provides the baseline questions for that comparison.

Path three: add radar for a defined supporting role

A complementary radar may provide information used for earlier navigation adjustments or another specified perception task while the established protective system remains responsible for its safety function. Define the additional behavior precisely. More sensor data is not an operational benefit until the controller uses it to improve a measurable result.

In a radar lidar sensor fusion proposal, ask what happens when the sensors disagree. The integration must preserve the authority of the protective function and handle missing, stale or conflicting data appropriately. An ordinary perception input must not silently override a protective stop because another sensor appears to see clear space.

This creates an important limit on the business case: if the retained laser scanner still reacts to dust with a protective stop, adding an ordinary radar does not by itself remove that stop. The additional sensor may improve a different task, but the original availability problem remains until an approved design change addresses its cause. Ask the supplier to identify the exact mechanism behind any claimed reduction in interruptions. Otherwise, the trial can show better perception without demonstrating the operational benefit that justified the purchase.

Path four: replace only after the responsibility transfer is complete

This path requires the clearest contract. Identify the removed hardware, the functions being reassigned, the new safety architecture, the remaining navigation sensors and the required validation. Record who owns integration and who approves release. Avoid a responsibility gap between a component vendor, robot manufacturer and site integrator.

Where the change affects response behavior or protective coverage, update the relevant stopping and field evidence. Our protective-field engineering guide addresses that work. Radar's resistance to a particular disturbance does not independently justify a smaller field, higher speed or shorter stopping margin.

Make the Supplier Demonstration Answer a Purchasing Question

For AMR radar sensor selection, start with a written hypothesis: this proposed change should reduce a specific class of avoidable interruption while preserving the required protective and navigation performance. The hypothesis should identify the route, environment, payloads and operating modes to which it applies.

Compare alternatives using consistent exposure. Record completed cycles and time spent in the relevant conditions, rather than reporting only total stops. Separate correct protective interventions, sensing-related unnecessary stops, navigation failures and unrelated delays. A supplier should not receive credit for reducing stops simply because fewer people crossed the route during its demonstration.

For each important event, retain enough information to connect the scene to the controller's response: selected mode, relevant sensor state, vehicle state and outcome. Where timestamps from different systems are used, establish how they can be compared. A visually smooth playback is less useful than a traceable explanation of an unexpected stop or missed requirement.

Request answers to these seven questions before agreeing that the trial demonstrates replacement readiness:

1. Which functions were demonstrated, and which requested functions remain untested?
2. Was the sensor stationary or moving, and were the proposed vehicle motions represented?
3. Which human and material target conditions were covered by the applicable evidence?
4. What happened when a target stopped, a field changed or sensor information became unavailable?
5. Were the normal payload, covers and installed mounting arrangement present?
6. Which operating restrictions remain, and what do they cost in throughput or supervision?
7. Who owns unresolved deviations and the final release decision?

Personnel-protection challenges belong in a controlled test arrangement using the appropriate procedures and test equipment. A productive trial does not require exposing workers to an unvalidated moving vehicle.

Keep the outcome specific. A trial may support adding radar for one loading-area task while providing insufficient evidence to remove a scanner across the whole site. That is still a useful purchasing result if the approved scope and expected value are clear.

Calculate the Cost of the Installed Decision

Sensor unit price is only one part of the comparison. Include controllers, cables, brackets, approved covers, engineering, software integration, validation, documentation, spares and training. Also include any navigation hardware that must remain after the proposed safety upgrade.

Estimate benefits from verified avoidable losses. Useful categories include unnecessary interruption time, associated operator intervention and maintenance effort that the new design can actually reduce. Avoid counting the same recovered time twice as both production value and labor saving unless the accounting basis supports both.

Consider a deliberately simple, hypothetical example. Assume the installed change costs $12,000. A comparable trial supports an estimate of 40 hours of recoverable interruption time per year, valued at $120 per hour, plus a separate $2,000 annual maintenance benefit. Assume the new equipment adds $800 per year in support and upkeep. Annual net benefit would be $6,000: $4,800 plus $2,000 minus $800. Simple payback would be two years.

If the recoverable interruption time is only 20 hours, annual net benefit falls to $3,600 and simple payback becomes approximately 3.3 years. These figures are arithmetic examples, not market prices or promised savings. The exercise shows why a trustworthy stop classification matters as much as the component quotation.

There can also be a capacity constraint. Recovered AMR time has limited financial value if a downstream station remains the bottleneck and the same output would be produced anyway. The business case should identify whether the gain changes shipped output, staffing needs, overtime or service reliability.

Finally, compare the rollout burden. A design that can be supported across a fleet may be more valuable than a slightly better demonstration requiring unique tuning for every vehicle. Ask how configurations are backed up, how replacement units are commissioned and what must be rechecked after firmware, mounting or payload changes.