SCARA arms cycle in 0.3-0.5 s on a 25/305/25 mm benchmark stroke, and that throughput is what locks them into electronics, screw-driving, and small-parts palletising cells where a 6-axis cobot would be slower by a factor of 2-3 [S4].
Collaborative robots sacrifice that raw speed and joint stiffness for power-and-force-limiting (PFL) joints, ISO/TS 15066-compliant contact limits, and a 1 m reach that lets a single operator hand-guide or share the workspace without light curtains [S1].
Kinematic Architecture and Working Envelope
SCARA stands for Selective Compliance Assembly Robot Arm: two parallel-axis rotary joints (J1, J2) handle the planar XY motion, a prismatic Z joint adds 50-200 mm of vertical stroke, and a final rotational joint (J4) orients the end-effector — four axes in total, with a 3-4 kg sweet spot at 0.3-0.6 m radius [S3]. A SCARA is selectively compliant in the X-Y plane and rigid in Z, which is exactly why it excels at vertical insertion and screw-fastening tasks where off-axis forces are absent [S2].
Cobots are six-axis articulated arms (J1-J6) with a serial kinematic chain: shoulder, elbow, three wrist joints producing full 6-DOF pose, payload classes tiered at 3-5 kg (UR3e), 6-10 kg (UR10e, Doosan A0509), 12-16 kg (FANUC CR-15iA), and up to 30-35 kg (KUKA LBR iisy, Doosan H2515) [S1][S4]. The articulated reach typically spans 0.5-1.3 m, and because the six joints decouple rotation from translation, the working envelope is closer to a 2/3 sphere than the disc that defines a SCARA's workspace [S4].
Repeatability, Speed, and Cycle Time
SCARA repeatability sits in the ±0.01-0.03 mm band versus ±0.02-0.1 mm for a six-axis cobot, and that 3-10x precision gap is the dominant selection lever in connector-insertion or micro-screwdriving cells [S4]. SCARA cycle time on a 25/305/25 mm stroke (vertical 25 mm up, 305 mm horizontal, 25 mm down) lands at 0.3-0.5 s for tier-1 units such as the Epson LS6-B, Yamaha YK-XG, and ABB IRB 910SC, while a comparable cobot cycle is 0.8-1.5 s [S4].
SCARA maximum joint speed reaches 7,000-10,000 mm/s of composite TCP velocity, with the limiting factor being the prismatic Z axis on screw-driving strokes; cobot TCP speeds top out at 2,000-3,000 mm/s, constrained by the PFL algorithm and joint torque sensor bandwidth needed for collision detection under ISO/TS 15066 [S4]. For a hard-throughput target — say 60 picks per minute on a 50 mm indexed tray — only the SCARA architecture has the margin.
Payload, Reach, and Mechanical Footprint
SCARA payload classes break cleanly into 1-3 kg, 3-6 kg, 6-12 kg, and 12-20 kg tiers, with reach options of 120 mm, 180 mm, 225 mm, 350 mm, 400 mm, 500 mm, 600 mm, 700 mm, 800 mm, and 850 mm; the 350-600 mm reach is the dominant share in Asian consumer-electronics lines [S4]. A SCARA pedestal footprint is roughly 200 x 200 mm for a 6 kg unit, and the arm never extends below the base plane — an advantage on top-mount conveyor cells where overhead clearance is limited.
Cobot pedestal footprint is 150-200 mm diameter for the 6-10 kg class, but the swept sphere of the wrist means at least 1.2-1.5 m of radial clearance is required for full reach; obstacle-avoidance paths in SCARA vs articulated robot comparisons consistently show SCARA swept volume at 30-50% of an equivalent cobot envelope at the same 600 mm radius. Cobots win on absolute payload ceiling (30-35 kg vs 20 kg SCARA) and on horizontal reach above 800 mm, where most SCARAs require the next reach tier or a linear track add-on.
Safety, Standards, and Human-Robot Collaboration
ISO/TS 15066:2016 specifies the biomechanical limits — quasi-static contact 50 N on the hand, 140 N on the chest, transient 320 N on the palm — that a PFL cobot controller must respect, and ISO 10218-1/-2 cover the broader industrial-robot safety requirements that both classes meet by default [S1]. A SCARA in power-and-force-limited mode (the so-called "cooperative SCARA" class) is rated below these limits only at reduced speeds under 250 mm/s TCP; outside that envelope it reverts to ISO 10218-1 safeguarding (light curtains, safety-rated scanners, hard guarding) [S1].
Practical floor-level consequence: a SCARA cell in a 120 ppm electronics line still needs physical guarding because cycle pressures exceed ISO/TS 15066 limits, while a 6-axis cobot loading a 5 kg CNC fixture can be specified fenceless with power-and-force limiting plus a speed-and-separation monitor (SSM) per ISO 13849-1 PLd. Cobots also dominate when the articulated robot footprint rules out a ceiling-mount gantry and the cell has to share operators on the same side of the conveyor.
Programming, Integration, and Ecosystem
Cobot vendors (Universal Robots, FANUC, KUKA, Doosan, Techman) ship hand-guiding UIs, drag-and-drop teach pendants, and a default I/O stack on EtherNet/IP, PROFINET, and Modbus TCP — the collaborative robot ecosystem now spans over 200 certified UR+ end-effector kits, and the open ROS 2 driver layer on the UR e-Series has accelerated research projects like the SCARA_ROBOT_ROS2 simulation stack [S2].
SCARAs from Epson, Yamaha, ABB, Mitsubishi, and Stäubli ship with vendor-proprietary motion controllers plus increasingly open OPC UA and ROS 2 drivers; field integration is dominated by ladder logic on a PLC rather than a tablet UI, and the buyer-side team usually includes a SCARA-certified application engineer rather than an end-user "I taught it in 20 minutes" workflow [S4]. The integration cost gap is roughly 2-4x in man-days for a SCARA cell versus a cobot cell at equal payload, and that delta is one of the cobot's strongest commercial arguments.
Decision Matrix: Pick by Use-Case
For planar assembly under 6 kg at 60+ ppm: SCARA wins on cycle, precision, and footprint — typical Epson LS6-B402S 400 mm / 6 kg class at 0.4 s cycle, ±0.01 mm repeatability. For full 6-DOF pose tasks (deburring, multi-face inspection, dispensing around curved surfaces) under 16 kg: cobot wins — the 6-axis kinematic chain is structurally required. [S2]
For medium-payload machine tending of 15-35 kg workpieces: cobot wins on payload and human-shared loading. For small-batch high-mix production with redeployments: cobot wins on teach-and-replay. For 24/7 dedicated high-speed lines: SCARA wins on cost-per-cycle and mean-time-between-failure. For washdown or cleanroom environments, stainless SCARA enclosures (IP65/67) are more common than full stainless cobot arms, which mostly exist in the 3-5 kg class.
Limits, Failure Modes, and Common Spec Traps
The two largest selection errors in the field are treating "collaborative" as a property of the robot alone (it is a cell-level safety function, not a unit-level feature) and underrating a SCARA's off-axis moment load — most 6 kg SCARAs lose 50% of rated payload when the tool centre is offset 30 mm from the J4 axis [S4]. Watch also for: cobot payload ratings at full speed (Doosan A0912 is 9 kg at low speed, derated to 5 kg at 1 m/s TCP), and SCARA cycle-time numbers quoted at a 2 kg payload that fall by 20-30% at 6 kg.
Another trap: cycle-time comparisons on different strokes. A SCARA cycle of 0.30 s at 25/25/25 mm does not extrapolate to a 25/305/25 mm stroke — that number doubles. Always ask vendors to quote the stroke you actually run, and always verify PFL contact-limit data with a robot-mounted force sensor rather than trusting marketing spec sheets alone.
Sourcing, Standards, and What to Verify on the Datasheet
The standards stack to check before purchase: ISO 9283 (manipulating industrial robots — performance criteria and test methods) for repeatability and accuracy numbers, ISO/TS 15066 for collaborative-mode limits, ISO 10218-1/-2 for general robot safety, ISO 13849-1 for the safety-rated parts of the control system, and IEC 61000-6-2 / IEC 61000-6-4 for EMC compliance on the cabinet [S1]. For PFL-mode contact force, request the vendor's biomechanical exposure data sheet rather than a generic "safe" stamp.
For spec-first cell sizing on adjacent motion subsystems, see the picking-by-positioning-accuracy linear module guide and the AGV/AMR path planning overlap for material-flow handoffs.