How Robotic Arms Work: Joints, Axes, Motors, and Control Explained
How robotic arms work — degrees of freedom and axes, the joints and servo motors, kinematics, the controller, end effectors, and how arms are programmed.
A robotic arm is a reprogrammable mechanical manipulator that mimics the structure and range of motion of a human arm. It consists of rigid links connected by motorized joints, driven by servo motors through a controller that coordinates every axis simultaneously. The controller reads encoder feedback from each joint hundreds of times per second and adjusts motor torques to keep the arm on its programmed path — whether that path is a weld seam, a pick-and-place cycle, or a precision assembly operation.
That is the short answer. The longer answer — covering anatomy, axes, kinematics, drives, and how a robot cell ties to a PLC — is what this guide is for.
What Is a Robotic Arm?
A robotic arm is a type of industrial manipulator designed to move a tool or workpiece through three-dimensional space with repeatable accuracy. Unlike a fixed machine that performs one motion, a robotic arm can be reprogrammed to perform different tasks by changing its program — the same mechanical hardware can weld on Monday and apply adhesive on Tuesday.
The defining characteristic of a robotic arm is its kinematic chain: a series of links and joints arranged so that the combination of joint rotations or translations produces arbitrary tool-tip positions and orientations within a defined workspace. Most industrial arms are serial manipulators — joints are arranged end-to-end — though parallel manipulators (delta robots) arrange joints in a closed-loop structure for higher speed at the cost of reduced reach.
Robotic arms are used across virtually every manufacturing sector: welding, painting, machine tending, assembly, inspection, palletizing, and material handling. Understanding how they work mechanically and electrically is essential for anyone working in automation engineering, PLC programming, or robot integration.
The Anatomy of a Robotic Arm
Every industrial robot arm shares the same structural hierarchy regardless of brand. Understanding each component clarifies both the mechanics and the control strategy.
Base
The base is the fixed foundation that anchors the arm to the floor, ceiling, or gantry structure. It houses the first joint and, in many designs, contains cable management passages for motor power and signal wiring. The base must absorb the reaction forces generated when the arm accelerates heavy payloads, which is why robot bases are typically cast iron or heavy-gauge steel bolted directly to a concrete foundation.
Links
Links are the rigid structural members that connect one joint to the next. They transmit motion without deforming. Link length and geometry determine the robot's reach envelope and influence the torque demands placed on each joint motor. Longer links require motors to overcome greater lever-arm forces, which is why robots with large reach envelopes use heavier servo systems at the shoulder and elbow joints.
Joints
Joints are the actuated connections between links. Almost all industrial robot arm joints are revolute joints — they rotate around a single axis. A small number of robot designs also include prismatic joints, which translate linearly along an axis (common in Cartesian and SCARA configurations).
Each revolute joint adds one rotational degree of freedom to the arm. The joint contains the servo motor, a gearbox or harmonic drive to multiply torque, and an encoder or resolver to feed position data back to the controller.
Wrist
The wrist is the cluster of joints at the end of the arm closest to the tool. A standard six-axis robot has a spherical wrist — three intersecting revolute axes whose intersection point stays fixed as the wrist rotates. This geometry simplifies the inverse kinematics calculation because wrist orientation can be solved independently from arm position.
End Effector
The end effector (or end-of-arm tooling, EOAT) is the device that interacts with the workpiece. It mounts to the robot's wrist flange and is not part of the robot itself — it is application-specific tooling that can be swapped between jobs. Common end effectors include pneumatic parallel grippers, vacuum suction cups, welding torches, and deburring spindles. For a detailed comparison of end effector types, see our guide on robot gripper types.
Degrees of Freedom and the 6 Axes
Degrees of freedom (DOF) is the number of independent motions a robot can perform. To position a rigid body arbitrarily in three-dimensional space — at any point in its workspace, at any orientation — requires six degrees of freedom: three for position (X, Y, Z) and three for orientation (roll, pitch, yaw).
This is why the 6-axis articulated arm is the dominant configuration in industrial automation. Six revolute joints give the robot the freedom to place its tool tip at a target point with a specific approach angle, which is critical for tasks like welding (where the torch must approach at the correct angle to the seam) or assembly (where a part must be inserted along a defined axis).
The Standard 6 Axes
| Axis | Common Name | Motion | Body Part Analogy |
|---|---|---|---|
| J1 (Axis 1) | Waist / Rotation | Rotates the entire arm around the base | Waist |
| J2 (Axis 2) | Lower arm / Shoulder | Pivots the lower arm forward and backward | Shoulder |
| J3 (Axis 3) | Upper arm / Elbow | Pivots the upper arm up and down | Elbow |
| J4 (Axis 4) | Wrist roll | Rotates the wrist assembly around its own axis | Forearm rotation |
| J5 (Axis 5) | Wrist bend / Pitch | Tilts the wrist up and down | Wrist bend |
| J6 (Axis 6) | Wrist twist / Roll | Rotates the tool flange | Wrist twist |
Axes J1–J3 determine where in space the wrist center is located. Axes J4–J6 determine how the tool is oriented at that point. This separation simplifies programming and control.
7-axis arms add a redundant joint — typically an additional "elbow swing" axis — to allow the robot to reach around obstacles. The extra DOF gives the arm multiple joint-space solutions for the same tool-tip position and orientation, which is useful in confined cells.
How Joints Move: Servo Motors, Gearboxes, and Encoders
Each robot joint is a closed-loop motion axis. Three hardware components work together to produce accurate, repeatable joint motion.
Servo Motors
Robot joints are driven by AC servo motors — brushless permanent magnet motors that deliver high torque density and respond rapidly to velocity commands. The motor itself provides raw torque; it does not inherently know its position or resist disturbances without the feedback loop.
For detailed coverage of how these motors work electrically, see our guide on servo motor control.
Gearboxes and Harmonic Drives
A servo motor alone cannot deliver the torque required to hold and move heavy links against gravity. Gearboxes multiply torque at the cost of speed. Industrial robots commonly use two gearbox types:
- Planetary gearboxes — compact, high-ratio, widely used at shoulder and elbow joints where torque demands are highest
- Harmonic drives (strain wave gearing) — zero backlash, very high reduction ratio in a compact package, commonly used at wrist axes where positional accuracy matters most
Zero backlash is critical. Even a small amount of gear play translates directly into tool-tip positioning error because the joint encoder sits on the motor shaft — it measures the motor's rotation, not the output shaft's position. Any backlash in the gearbox is invisible to the encoder.
Encoders
An encoder is a feedback sensor that converts shaft rotation into a digital position signal. Robot joints use high-resolution multi-turn absolute encoders — typically 17 to 23 bits per revolution — that report the exact angular position of the motor shaft every time the controller requests it.
Absolute encoders retain position data across power cycles. This is important: an industrial robot arm does not require a homing routine after a power cycle because each joint already knows its absolute position when the controller powers up.
For a side-by-side comparison of feedback sensor types, see the encoder vs resolver guide.
Kinematics: Forward and Inverse in Plain Terms
Kinematics is the branch of mechanics that describes motion without reference to forces. Robot kinematics deals with the relationship between joint angles and tool-tip position and orientation. There are two directions to this problem.
Forward Kinematics
Forward kinematics answers the question: given the current joint angles, where is the tool tip?
If you know J1 = 30°, J2 = 45°, J3 = −20°, J4 = 0°, J5 = 90°, J6 = 0°, where exactly is the tool center point in Cartesian space? Forward kinematics computes that answer using a chain of geometric transformations — one transformation per link — defined by the robot's physical geometry (link lengths, joint offsets, twist angles). The math produces a definitive answer with no ambiguity.
Forward kinematics is used by the controller continuously — every millisecond, the controller reads the current joint angles from the encoders, runs the forward kinematics calculation, and knows exactly where the tool tip is in the work cell coordinate frame.
Inverse Kinematics
Inverse kinematics (IK) answers the reverse question: given a target tool-tip position and orientation, what joint angles are required?
This is much harder. For a 6-axis arm, IK is not a single equation — it is a system of nonlinear equations with multiple valid solutions (called configurations or elbow configurations). A typical 6-axis robot with a spherical wrist has up to eight analytical solutions for any reachable Cartesian target. The controller must select the solution that:
- keeps the robot within joint limits
- avoids singularities (configurations where the robot loses a degree of freedom)
- minimizes joint travel from the previous position
When a programmer teaches a target point using a pendant and records its Cartesian coordinates, the controller runs IK at execution time to determine the joint angles needed to reach that point. This is transparent to the programmer, but understanding it explains why a robot can behave unexpectedly near singularities — mathematical configurations where the IK solution becomes undefined or requires infinite joint velocity.
Singularities occur at specific geometric configurations: wrist singularity (J5 ≈ 0°, where J4 and J6 become parallel), shoulder singularity (wrist center directly above the robot base), and elbow singularity (arm fully extended or retracted). Most robot controllers detect singularity proximity and either slow down or produce an error.
The Robot Controller
The robot controller is the industrial computer that orchestrates all six servo axes simultaneously. It is architecturally similar to a PLC but purpose-built for coordinated multi-axis motion. Its key functions are:
Trajectory planning — converting the programmed path into a sequence of interpolated set-points at the servo update rate (typically 1–4 ms). If the path is a straight Cartesian line, the controller continuously runs IK to transform each interpolated Cartesian point into joint angles.
Servo loop execution — for each axis, running a position/velocity/torque control loop at high update rates (typically 250 Hz to 1 kHz). The loop compares the encoder-measured position to the planned set-point and commands the servo drive to correct the error.
I/O management — reading digital and analog inputs from sensors and writing outputs to grippers, solenoids, and status signals. The controller's I/O system is how the robot interacts with fixtures, conveyors, and safety devices.
Safety monitoring — enforcing speed limits, workspace limits (software fences), and emergency stop logic. Modern controllers implement functional safety (IEC 62061 / ISO 13849) directly in the controller firmware rather than relying solely on external safety relays.
Servo Drives Per Axis
Each joint has a dedicated servo drive — a power electronics module that converts the DC bus voltage into variable-frequency three-phase AC at the commanded current. The drive receives a velocity or torque reference from the controller over a high-speed internal bus and closes a fast current loop (typically at 8–16 kHz) to follow that reference.
The architecture is: controller (trajectory + position loop) → servo drive (velocity + current loop) → servo motor → gearbox → joint. This cascaded structure allows the controller to manage six independent axes while each drive handles the fast inner loops locally.
For the broader context of how drives, motors, and controllers interact, see the motion control basics guide.
How the Robot Cell Ties to a PLC
The robot controller and the cell PLC are peers on the same industrial network. The PLC is the cell supervisor — it manages the production sequence, conveyors, part-present sensors, fixture clamping, and safety zone interlocks. The robot controller is the motion executor — it runs the robot program when the PLC tells it to.
Communication typically happens over PROFINET, EtherNet/IP, or DeviceNet, depending on the robot brand and the PLC platform. The PLC sends handshake signals to the robot controller — program number select, cycle start, emergency stop — and the robot controller sends back status signals — at home position, program running, fault, cycle complete. For applications requiring tight path coordination (like a robot synchronized to a moving conveyor), some controllers support conveyor tracking via a dedicated encoder input that the controller reads directly.
This division of responsibility is standard in industrial practice: the PLC owns the cell logic, the robot controller owns the motion. A PLC programmer integrating a robot for the first time will recognize the I/O handshake model immediately — it mirrors how a PLC controls any other specialized device.
End Effectors and Grippers
The end effector defines what the robot can do with a workpiece. It is mounted to the J6 wrist flange and must be mechanically and electrically compatible with the robot's payload rating and wrist interface (ISO 9283 flange standards define the bolt pattern and alignment pin geometry).
Common end effector categories include:
- Pneumatic parallel grippers — two or three jaws driven by an air cylinder; used for rigid parts with consistent geometry
- Vacuum suction cups — silicone or urethane cups connected to a venturi or vacuum pump; used for flat, smooth, or delicate surfaces
- Magnetic grippers — for ferrous sheet metal where suction is impractical
- Welding torches — part of the robot's tool (TCP) definition; the wire feed and gas solenoid are controlled via the robot I/O
- Force-torque sensors — mounted between the wrist flange and the tool to give the robot tactile feedback; used in assembly tasks requiring compliant insertion
For a detailed breakdown of gripper types, mechanisms, and selection criteria, see the full guide on robot gripper types.
Tool Center Point (TCP)
The tool center point is the virtual reference point that the controller tracks through space. For a welding torch it is the wire tip; for a suction gripper it is the center of the cup face. TCP calibration — teaching the controller the geometric offset from the wrist flange to the actual tool tip — is a mandatory commissioning step. An inaccurate TCP causes the robot to arc through the air with the wrong point as its reference, producing path errors that look like incorrect joint angles but are actually a calibration issue.
How Robotic Arms Are Programmed
Robot arms are programmed by defining a sequence of target positions, motion types, speeds, and I/O actions. The most common programming methods are:
Teach pendant (online programming) — the engineer jogs each joint individually or in Cartesian mode to move the robot to a target position, then records that position as a program point. The pendant displays joint angles, Cartesian coordinates, and I/O states. Teach pendant programming is the default method for initial commissioning and for programs with a moderate number of points.
Offline programming (OLP) — the robot program is written in a simulation environment using a 3D model of the cell. The programmer defines paths, orientations, and logic on a desktop, then downloads the program to the real robot. OLP is used for arc welding paths, laser cutting, and any application where path accuracy requires hundreds of programmed points that would be impractical to teach manually.
Lead-through programming — the operator physically grasps the robot arm and moves it through the desired path; the controller records joint positions at set intervals. Used for spray painting and other applications where the skilled operator's natural motion produces the best path.
For a comprehensive walkthrough of all programming methods, robot languages (FANUC TP/Karel, ABB RAPID, KUKA KRL, Yaskawa Inform), and step-by-step program creation, see the industrial robot programming complete guide.
Programming a robot for pallet stacking involves additional considerations — grid pattern generation, layer logic, and conveyor handshake — covered in the robot palletizing guide.
Frequently Asked Questions
How does a robotic arm work?
A robotic arm works by using servo motors at each joint to rotate rigid links through programmed angles. A controller runs inverse kinematics to convert Cartesian target positions into the joint angles required to reach them, then commands each axis's servo drive to move the joint to that angle while reading encoder feedback to confirm the position was reached. The result is a tool tip that moves through three-dimensional space with high repeatability.
What are the 6 axes of a robot?
The 6 axes of a standard industrial robot arm are: Axis 1 (waist rotation), Axis 2 (shoulder forward/backward), Axis 3 (elbow up/down), Axis 4 (wrist roll), Axis 5 (wrist bend/pitch), and Axis 6 (wrist twist). Axes 1–3 position the wrist center in space; axes 4–6 orient the tool at that position.
What is the difference between forward and inverse kinematics?
Forward kinematics calculates the tool-tip position and orientation from known joint angles — it is straightforward and has a single answer. Inverse kinematics calculates the joint angles required to place the tool at a given Cartesian position and orientation — it is computationally harder and may have multiple valid solutions. The robot controller uses forward kinematics continuously to track tool position and inverse kinematics during path execution to convert Cartesian targets into joint commands.
How are robotic arms controlled?
Robotic arms are controlled by a robot controller that runs trajectory planning and position loops for all six axes simultaneously. Each axis has a dedicated servo drive that executes inner velocity and current loops. The controller reads high-resolution absolute encoder feedback from each joint to close the position loop. In a production cell, a PLC supervises the robot controller via industrial network I/O — sending cycle start and program select signals and receiving status feedback — while the robot controller handles all motion execution internally.
