A single 2D safety LiDAR on a typical AMR is mounted about 200 mm (8 in) above the floor, scans a horizontal plane near the chassis, and costs roughly $1,500 to $5,000, while full 3D safety-rated systems run about $12,000 [S4]. That price and geometry gap is the reason fleets are layering 3D sensors or 3D ultrasonic ADAR on top of the 2D scanner for ISO 3691-4 / IEC 61508 safety functions [S1][S4].
Field of view on a safety LiDAR is the single biggest determinant of whether an AMR robot can detect a foot, a forklift fork, or a raised pallet, and 2025-2026 product data shows a clear split between 2D plane scanners, forward 3D blocks, and full hemispheric 3D sensors [S2][S3][S5].
2D safety LiDAR: 270°–360° in one plane, but blind above and below the scan line
A 2D safety LiDAR specified for AMR obstacle detection typically delivers 270° to 360° horizontal coverage in a single plane, with a safety range of roughly 1.5 m to 5 m at the rated reflectivity (white target, 10% reflectivity worst case) [S4]. The scanner's plane sits near 200 mm above the floor, so it will not see a 100 mm low obstacle or a 2 m overhang unless a second sensor covers that band [S1].
This geometry is exactly why several near-miss incident classes keep recurring: a person kneeling, a pallet jack tine left on the floor, a forklift with raised forks blocking an aisle [S1]. One 180°×180° ADAR sensor paired with a 2D safety scanner claims 3D 360° coverage at 16.4 ft (5 m) range for the safety function, and is positioned by its vendor as a direct response to that blind band [S1].
Forward 3D LiDAR: 90°–120° horizontal FOV with 90° vertical, for direction-aware AMRs
Forward-facing 3D LiDARs aimed at mobile robotics are commonly specified at 90° to 120° horizontal FOV and 90° vertical FOV, with 75 m reach on the automotive-grade E platform and 144 structured beams through a flat, flush window [S2]. The 90° vertical band is the practical minimum for catching both floor-level obstacles and waist-height protrusions in a single forward cone.
For an AGV robot or AMR running fixed routes, a forward 3D block is often the cost-effective middle ground: it covers the travel direction and a useful underhang/overhang band, but it cannot replace a 360° perimeter sensor for reversing, cross-traffic, or dynamic human zones [S3]. Beam count and points-per-second, not just FOV, drive whether the sensor can resolve a thin forklift fork at 5 m, a 426 KB cloud per frame at 10 Hz from a 16-channel unit scaling up to 5.2 million points per second for a 128-channel unit [S3].
Hemispheric 3D LiDAR: 360° × 90°+ for reverse, rotate, and dense-aisle work
Hemispheric 3D LiDARs built for mobile robots give a 360° horizontal FOV with 90°+ vertical, exemplified by the RoboSense Safety Airy family at 96 channels and the Livox Mid-360S, which delivers 360° coverage in a compact body aimed at AMR and AGV retrofits [S3][S5]. Pricing on these units is well below the 3D safety-scanner tier: open-market listings put the Mid-360S at roughly €827 to €861, with the Airy 96-channel at about €114 to €1,083 depending on configuration [S5].
The 360° view matters during the exact AMR motions a forward sensor cannot cover: rotation, reverse, entering narrow aisles, and navigating around dynamic obstacles, all of which the Mid-360S guide flags as core use cases [S5]. Vertical FOV in these units also makes them a credible single-sensor replacement for a 2D safety LiDAR plus a 3D camera fusion stack, at least on paper, with the trade-off that not all hemispheric sensors carry a safety certification to PLd / SIL 2, so a certified 2D scanner is often retained for the safety-rated field set [S3][S6].
Range classes, weight, and why a 200 m automotive unit is usually the wrong pick
Safety-rated 2D LiDARs land in the 1.5 m to 5 m safety range, near-field 3D units in the 5 m to 30 m range, and automotive-grade LiDARs at 75 m to 200 m+ for highway work, but a slow indoor AMR rarely benefits past 30 m and pays a heavy weight and compute penalty for the extra reach [S3][S4]. The Digiflec guide calls this out directly: "a slow indoor AMR doesn't need automotive-grade 200 m range" [S3].
Weight and power are the quiet budget killers. A 30 W, 1 kg 3D LiDAR is irrelevant on a 500 kg AGV but disqualifying on a small mobile robot or drone platform, so the FOV × range decision has to be weighed against the chassis's actual payload and battery budget before any range number is locked in [S3].
Sensor comparison: 2D LiDAR vs 3D LiDAR vs 3D camera vs ADAR, on four decision axes
Across the four sensors most often pitched for AMR safety, the trade space lines up clearly: 2D safety LiDAR gives certified 360° in one plane at $1,500 to $5,000 but misses underhangs and overhangs; 3D safety-rated LiDAR extends to full 3D around $12,000 with FOV from 90° to 360° [S4]. 3D cameras add rich visual context and overhang detection, are not safety certified, and degrade with reflections and variable lighting [S1]. 3D ultrasonic ADAR claims 180°×180° at 5 m as a 3D safety-grade alternative targeting the underhang/overhang gap left by 2D LiDAR [S1].
On the four decision axes that matter for AMR procurement, the matrix is: (1) FOV coverage, 2D LiDAR is 270° to 360° in a plane, 3D LiDAR 90°×90° to 360°×90°+, 3D camera up to 360°×180° at short range, ADAR 180°×180° per unit; (2) safety certification, only 2D and some 3D LiDAR plus ADAR carry PLd / SIL 2 claims, cameras do not [S1][S3][S4]; (3) blind-spot class, 2D LiDAR misses above/below plane, 3D camera misses in low light or against retroreflective vests, 3D LiDAR can miss transparent and small cross-sections under roughly 30 mm at distance; (4) cost band, $1,500 to $5,000 for 2D safety, about $12,000 for a full 3D safety system, $800 to €1,000 for non-safety hemispheric 3D LiDAR, and ADAR pricing sits between the 2D and full 3D safety tiers per its vendor [S1][S4][S5].
Use cases by FOV: picking the right geometry for the aisle
Indoor AMRs and AGVs on fixed routes with mostly predictable cross-traffic are well served by a 2D safety LiDAR at 200 mm height plus a forward 3D camera for logging, which is the most common 2025-2026 stack and the one conveyor sorting line handoffs into the safety field rely on [S1][S6]. Mixed-traffic warehouses with forklifts, raised forks, and human pickers need a hemispheric 3D LiDAR or a 2D scanner + 3D safety-rated sensor to cover the 200 mm to 2 000 mm band that a single 2D scanner leaves open [S1][S5].
Outdoor mobile robots moving above roughly 1.5 m/s, such as autonomous inspection and last-mile units, benefit from forward 3D LiDAR with 75 m class reach and 90°+ vertical FOV, paired with short-range 2D safety scanners for the close-in field where a 75 m beam is overkill and creates too much data [S2][S3]. Cobot arms and warehouse dimensioning systems need neither; there, accuracy and precision outrank FOV, and the FOV-driven geometry that matters for a 200 kg AMR becomes a poor fit on a 10 kg arm.
Limitations, failure modes, and the standards that frame them
2D safety LiDAR's largest failure mode is geometric: a person kneeling at 400 mm height and a 100 mm low pallet are invisible to a scanner mounted at 200 mm unless a second sensor covers the lower band, and 3D cameras fail under retroreflective PPE, glare, and featureless walls where SLAM drifts [S1][S3]. Hemispheric 3D LiDAR solves coverage but introduces compute load, with a 128-channel unit pushing roughly 5.2 million points per second into the AMR's onboard pipeline [S3].
Standards framing: AMR safety in Europe is anchored on EN ISO 3691-4 (driverless industrial trucks) and the functional-safety stack underneath it (IEC 61508, ISO 13849-1), while 3D ultrasonic ADAR vendors are explicitly positioning against the 2D-LiDAR blind band that 3691-4 compliance is being audited against [S1]. Pricing on safety-rated 2D scanners has compressed into the $1,500 to $5,000 band, against roughly $12,000 for a full 3D safety system, a 2.4× to 8× delta that procurement teams are weighing per unit [S4].
Sourcing and what to track next
Trackable signals: 2026-09 buyer-guide snapshots list RoboSense Safety Airy as a 3D scanner built to the same certification standards as 2D safety LiDAR with a 90° vertical FOV, an indicator that 3D safety-grade LiDAR is crossing from automotive to mobile robotics at price points below the legacy $12,000 tier [S3]. The 2026-07 Mid-360S guide and 2026-06 vendor market reports both flag falling 360° 3D LiDAR pricing as the main disruptor of the 2D safety LiDAR + 3D camera stack [S5][S7].
Near-field perception research published 2025-12 confirms that conventional ranging sensors leave systematic blind zones for AMRs in manufacturing, which lines up with the same underhang/overhang problem 2D LiDAR has had for a decade, and is the engineering case the new 3D and ADAR sensors are being bought to solve [S8]. The 2025-12 Plant Engineering piece closes with a usable rule: a foot at the floor needs roughly 300 mm of added safety distance around a 200 mm-mount 2D scanner, before any speed or stopping-braking math is layered on [S1].