For most 2026 buyers, a six-axis articulated robot in the 3–270 kg payload range — TM Robotics TVL500 (3 kg / 0.6 m reach) at the light end, ABB IRB 6730S (130–270 kg / 3.1–4.0 m reach) at the heavy end — covers roughly nine out of ten discrete material-handling, machine-tending and assembly cells [S5][S2].
The right machine is dictated by four hard spec gates: payload, reach, repeatability and the IP rating of the wrist/forearm, with axis 4–6 torque and mounting style (floor vs. ceiling) as the second-tier levers. Efort ER12-900 (6-axis, assembly/machine-tending) and Comau NJ-370-2.7 (6-axis, 50 kg forearm auxiliary load, 2109 Nm on axes 4–5) are the two middle-band reference points buyers should benchmark against [S1][S4].
Payload and Reach: The Two Numbers That Gate Everything Else
Payload divides the 2026 articulated-robot market into four working bands, each with a different vendor cluster [S1][S2][S4][S5]. Micro/light (≤5 kg) is dominated by compact ceiling-mountable units such as the TM Robotics TVL500 (3 kg payload, 602 mm reach, 0.02 mm repeatability, 8 m/s composite speed, 28 kg machine weight) — designed for food, electronics, pharmaceutical and clean-room cells [S5]. The 5–20 kg band is where Efort ER12-900 sits as a 6-axis assembly/tending platform [S1]. Medium 100–200 kg covers Comau NJ-370-2.7 (with 50 kg additional forearm auxiliary load and 2109 Nm torque on axes 4 and 5, axis 6 at 1177 Nm) and ABB IRB 6620 (150 kg / 2.2 m) [S4][S2]. Heavy 200+ kg is ABB territory with the IRB 6730S at 130–270 kg payload and 3.1–4.0 m reach [S2].
Reach is not linearly scalable with payload — the ABB IRB 1410 (5 kg) gives 1.44 m of reach at 5 kg, while the IRB 6730S stretches 4.0 m at up to 270 kg, meaning a 54× payload increase buys a 2.8× reach increase [S2]. Buyers should always verify the payload-vs-reach envelope at the wrist center, not at the faceplate, because suppliers quote reach to the centerline of axis 5, not to the tool tip.
Repeatability, Speed, and Axis Torque: The Hidden Specs
Repeatability in 2026 articulated robots is commodity-grade at the 6-axis light/medium end: the TVL500 is rated at 0.02 mm (0.0008 in) and the IRB 1410 class runs similar — adequate for dispensing, pick-and-place, light arc welding, and machine tending where path accuracy is process-driven, not robot-driven [S5][S2]. For sub-0.01 mm applications (laser cutting, micro-assembly) buyers must move to a SCARA or a custom linear stage, not escalate the articulated robot budget.
Axis torque on joints 4, 5, 6 — the wrist triplet — is the figure that decides whether a heavy gripper, servo-driven tool changer, or vision camera can be mounted without overheating the wrist during a continuous duty cycle. Comau NJ-370-2.7 publishes 2109 Nm on axes 4 and 5 with 1177 Nm on axis 6 and axis-1 stroke of ±180° at 85°/s, axis-2 stroke of +75°/-60° at 85°/s, axis-3 stroke of -10°/-231° at 85°/s, and axes 4–5 at 90°/s [S4]. ABB's mid-range IRB 6620 (150 kg, 2.2 m) and heavy IRB 6730S (130–270 kg, 3.1–4.0 m) sit above that envelope for high-inertia tooling [S2].
Mounting, Footprint, and Cleanroom/Washdown Specs
Mounting style is a buy-time decision, not a retrofit: the TVL500 is explicitly offered as floor-mounted or ceiling-mounted, a dual-mount configuration that doubles the cell-layout options for compact lines [S5]. Efort ER12-900 and Comau NJ-370-2.7 are 6-axis platforms described for assembly, machine tending, packaging and similar duty, with Comau listing axis-3 stroke of -10°/-231° — a useful inverted-mount or top-load geometry [S1][S4].
For cleanroom or washdown duty, the TVL500 specification sheet explicitly lists food, electronics, pharmaceutical, plastics and clean-room applications as target domains — a vendor that has validated the IP rating and surface finish for those cells, not just declared it [S5]. Buyers specifying a washdown or ISO Class 5+ cleanroom cell should request the vendor's IP65/IP67 wrist certification and cleanroom material certificate at the quote stage, not assume it from a generic 6-axis datasheet.
Mobile and Compound Articulated Robots: When the Arm Leaves the Pedestal
The 2026 product category has matured beyond the pedestal: RealMan's RMC-LA compound lifting robot pairs a six-axis arm with a lifting column and an integrated open-source 3D visual sensor at the end-effector, with positioning accuracy rated at < ±5 cm (1 mm on the lifting axis) and DC 24 V working voltage from a single-phase 220 V ±10% / 50 Hz supply, plus laser-scanning obstacle-avoidance as a standard safety function [S3]. The RMC-LA targets mobile-pick, tote-handling and warehouse-line cells where the workspace is larger than a 4 m pedestal reach can cover.
For fixed-installation cells, a 6-axis articulated robot is almost always the right answer over a Cartesian gantry when the working envelope is irregular, when the tool needs to approach the part from multiple angles, or when the work cell is shared between different parts. For a tall, narrow pick-and-place with a single fixed approach vector, a linear guide-based gantry will be cheaper and stiffer — buyers should not pay the articulated-robot tax for a geometry that does not need it.
Comparison: Four 2026 Reference Models Against Decision Criteria
The four most representative 2026 articulated robots in the public datasheets line up against the four buying criteria as follows. TVL500 (3 kg / 0.6 m / 0.02 mm / 28 kg) is the light/cleanroom pick. Efort ER12-900 (12 kg class / 6-axis / assembly & tending) is the small-parts mid-band pick [S1]. Comau NJ-370-2.7 (370 kg class / 6-axis / 50 kg forearm aux / 2109 Nm wrist / packaging) is the heavy-payload mid-band pick with strong wrist torque [S4]. ABB IRB 6730S (130–270 kg / 3.1–4.0 m / heavy welding & handling) is the heavy-large-envelope pick [S2].
Where the requirements are a 5 kg payload with a 1.44 m reach and arc-welding duty, the ABB IRB 1410 is the reference model [S2]. Where the requirement is a 150 kg payload at 2.2 m, the ABB IRB 6620 is the reference [S2]. Buyers who cannot map their cell to one of these published envelopes within ±10% on payload and ±15% on reach are almost certainly looking at a non-standard integration, and the vendor selection should be redone with a collaborative robot or a custom linear-axis hybrid in scope.
When an Articulated Robot Is the Wrong Tool
An articulated robot is the wrong machine when (a) the cycle time is dominated by long straight-axis travel with no reorientation, where a Cartesian or crossed-roller-guide gantry is faster and cheaper; (b) the payload exceeds 270 kg in a single 6-axis arm, where a 7th axis track or a heavy-payload gantry-hybrid becomes the only path; (c) the cell must operate in a certified collaborative mode with a human inside the workspace, where a purpose-built collaborative robot is the correct category rather than a guarded 6-axis arm with safety scanners added on. For mobile-pick in an open warehouse, an AGV or AMR base topped with a fixed arm is the standard 2026 pattern, not a free-standing pedestal robot. [S1]
Sources and What to Verify at Quote Stage
All product facts in this guide are drawn from vendor-published DirectIndustry listings dated 2026-05 to 2026-06: Efort ER12-900 [S1], ABB articulated-robot portfolio including IRB 1410/6620/6730S [S2], RealMan RMC-LA compound lifting robot [S3], Comau NJ-370-2.7 [S4], TM Robotics TVL500 [S5]. Patent landscape data on articulated-robot assembly was last surfaced on 2023-12-28 and is not a 2026 primary source [S6 (2023-12)].
At quote stage, buyers should request: (1) payload-vs-reach curve at the wrist center, not at the faceplate; (2) wrist IP rating (IP65/IP67) with vendor test certificate; (3) repeatability measured per ISO 9283, not vendor-claimed; (4) controller and fieldbus compatibility with the existing PLC line (PROFINET, EtherNet/IP, EtherCAT); and (5) a 24-hour continuous-duty thermal test for the wrist motors at full payload, since wrist-thermal-cut trips are the most common post-installation failure on heavy-payload cells.