AMR selection for electronics handling turns on four spec gates, not brand: chassis floor-load rating (typically 30-60 kg per wheel contact patch on raised-access floors), ESD-safe or anti-static wheel and deck materials, sensor-fusion navigation combining 2D LiDAR with vision, and a payload class matched to reel carriers, magazine racks, or tote stacks. The OTTO 100 (150 kg), OTTO 600 (600 kg), OTTO 1200 (1200 kg), OTTO 1500 (1900 kg), and OTTO Lifter (1200 kg) form a five-step payload ladder that aligns with the carrier sizes common in SMT line feed and back-end test [S2].
Electronics plants diverge from general material handling in three ways: light, static-sensitive payloads under 50 kg, narrow aisles between SMT lines, and the need to dock at conveyors, magazines, and AGV-era fixed routes. ABB's AMR brief [S1] frames AMRs as a step beyond fixed-path automated guided vehicles, with the autonomy layer mapping routes on the fly rather than following buried wire or magnetic tape, which is the lever electronics plants use to retrofit brownfield cleanrooms and back-end test halls without floor cutting.
Payload ladder and carrier fit
The single most consequential number on the data sheet is rated payload at the deck, not the marketing "max load". For electronics, three carrier weights dominate: 10-30 kg reel carts (typically 2-4 reels of 7-13 inch diameter), 50-150 kg magazine racks for SMT feeders, and 200-600 kg totes of finished boards on the back end. The 150 kg class (OTTO 100) covers light reel work, the 600 kg class (OTTO 600) covers magazine and small tote work, and the 1200-1900 kg class covers palletised finished goods [S2]. The AMR chassis must publish payload at a stated centre of gravity height, usually 600-900 mm for the smaller models and 700-1200 mm for the larger; exceeding this envelope derates the rated load and trips the fleet manager's safety limiter.
Total footprint matters as much as payload. Raised-access cleanroom tiles (commonly 600 x 600 mm) set a hard grid for wheelbase, and conveyors set docking tolerances of ±10-25 mm in X and Y. Choosing an AMR footprint that is an integer multiple of the tile grid avoids partial-tile load and the cracked-tile failure mode that drives most electronics-plant floor callbacks. See the broader AMR equipment taxonomy in the AMR robot reference for how wheel count, drive type, and chassis width map to aisle width.
ESD, cleanroom, and chassis materials
ESD protection is a chassis spec, not an accessory. Wheel material, deck surface resistivity, and grounding straps must keep the AMR body at the same potential as the conveyor or magazine it docks with, typically below 1 x 10^9 ohm surface resistivity for ANSI/ESD S20.20-compliant handling of class-1A or higher devices. Stainless or conductive-polymer wheels, copper-bonded ground straps, and a documented resistance-to-ground measurement at commissioning are the four checks that separate a qualified electronics AMR from a generic one. Pharma and food cleanrooms share this concern; the pharmaceutical AMR selection reference covers ISO 14644 class gating, and the food and beverage AMR selection reference covers washdown and stainless gates, both of which overlap with back-end test cleanliness. [S1]
Cleanroom class sets the IP rating and the blower/filter spec. ISO Class 7 (FED 209E Class 10,000) cleanrooms, common in semiconductor back-end, typically demand IP54 chassis and non-outgassing wheels; ISO Class 5 demands IP65 plus HEPA-filtered exhaust. Particles below 0.5 micron from the AMR drive train are the recurring root cause of post-AMR yield loss, and the spec to push back on the vendor is the documented particle count per metre travelled.
Navigation stack: 2D LiDAR plus vision fusion
Sensor fusion is no longer optional. The Springer review of 2D LiDAR plus vision fusion (YOLOv3 deep-learning detector feeding false laser scans into the ROS local cost map) demonstrated obstacle detection that 2D LiDAR alone misses, including low-contrast, low-profile, and transparent objects in the 0.05-0.3 m range that are exactly the dropped reels, tape strips, and stray cables of an SMT aisle [S3]. The architecture uses 2D LiDAR as the global SLAM sensor and vision as a complementary detector that injects virtual scans into the local cost map; this is a published, citable approach to closing the 2D LiDAR blind spot, and electronics plants should specify it in writing rather than accept a "vision-assisted" marketing line.
Vehicle dynamics matter when the aisle is tight. Mecanum four-wheel drive gives true holonomic motion in any direction, which the same Springer chapter validates, but it costs floor clearance and adds wheel scrub on contaminated tile. Differential drive is cheaper and more robust to metal debris, but it cannot crab sideways into a magazine dock. For most electronics AMR duty cycles the decision is between holonomic Mecanum for short, docking-heavy routes and differential drive for long, straight-line SMT line feed, with the mobile crane and articulated robot references only loosely related to the kinematics question and best left for the heavier-payload docks.
Fleet management and line-side integration
OTTO Fleet Manager is named as the fleet software that "gives complete control of material handling operations through one platform" [S2], and the integration points that electronics plants should gate on are: WMS/ERP REST or SQL hooks for work-order dispatch, OPC-UA or TCP socket bridges to the line-side PLC, and a documented API for magazine/conveyor handshake. The vendor-published metric of 13M+ production hours and 100+ AMR fleets in the field [S2] is the installed-base signal that drives the integrators' willingness to expose the API, and a fleet of 1M+ monthly deliveries is a useful proxy for the kind of throughput an electronics plant should expect once traffic is balanced.
AGV-to-AMR migration is the typical retrofit path. Plants that already run magnetic-tape or floor-wire AGVs will keep some fixed-route legs for SMT line feed, since fixed-path legs are more deterministic for takt-time-critical moves, and run AMRs on the open-floor back-end and warehouse legs. The AGV robot and storage handling references cover the upstream taxonomy; for electronics specifically the practical advice is to start the AMR pilot on the warehouse-to-line tote loop, not on in-line board transfer, and gate the in-line move only after the fleet software has logged at least 90 days of clean dispatcher behaviour.
Decision criteria: when to buy and when to wait
Buy an AMR when the route distance is over 50 m, the SKU count is over 200, or the existing AGV fleet is hitting its traffic-cap limit, per the intralogistics AMR review which frames the introduction pattern as "manufacturing, warehouses, cross-docks, terminals" [S4]. Do not buy an AMR for in-cell moves under 5 m, for under-50 kg single-piece flows that a conveyor handles cheaper, or for any move where the cycle time must be guaranteed to under ±2 s; AMRs optimise flexibility, not deterministic cycle time. The intralogistics literature also flags the planning problem as still open, with the Fragapane et al. (2021) paper in European Journal of Operational Research (vol. 294, no. 2, pp. 405-426) the cited planning-and-control reference [S4], which is the upstream of the 2024 chapter.
A direct comparison of the five OTTO payload classes on four decision criteria lines up as follows, with values taken from the published payload range and the typical electronics carrier weight in 2026. OTTO 100 (150 kg) maps to reel-cart and small-tote moves at lowest unit cost; OTTO 600 (600 kg) maps to magazine rack and mid-size tote moves with the most balanced aisle/payload trade; OTTO 1200 (1200 kg) and OTTO 1500 (1900 kg) map to pallet and finished-board moves where the route is open-floor; OTTO Lifter (1200 kg) is the deck-height-variable option for conveyor handoffs [S2]. For automotive parts and heavier-pallet duty cycles the automotive AMR spec gates reference covers the heavier end of the same fleet family.
Commissioning, validation, and failure modes
Three failure modes dominate the first 90 days: (1) wheel marking on raised-access tile from static-loaded parking, fixable with ESD wheel compounds and parking-pad tiles; (2) LiDAR false positives in glass-walled cleanroom corridors, fixable with vision-fusion overlay as the Springer chapter demonstrates [S3]; and (3) dispatcher deadlocks at conveyor junctions, fixable with a documented junction-priority policy and a PLC handshake that hands the right-of-way to the conveyor, not the AMR. Each of these is a known root cause in the storage handling failure literature, and a Site Acceptance Test (SAT) plan should reproduce all three before sign-off.
Trackable signals through 2026-08 include vendor-published fleet milestones (the OTTO reference of 13M+ production hours and 100+ fleets was published 2026-07-30 [S2]), peer-reviewed intralogistics planning updates, and the rollout of 3D LiDAR as a price-competitive alternative to 2D LiDAR plus vision fusion. The next node to watch is the next round of Springer/IEEE planning-and-control papers building on Fragapane et al., and any vendor release that publishes particle-count-per-metre data, since that single number will become the gating metric for ISO Class 5 electronics cleanroom deployments.