Quick answer The Denavit–Hartenberg convention describes each link by four parameters — twist α, link length a, offset d and joint angle θ — which build the transform ⁱ⁻¹Tᵢ between adjacent frames. Forward kinematics is their product T_0n = T_01·T_12·…·T_(n−1)n, giving end-effector pose directly.
Enter DH parameters (α, a, d, θ) to compute forward kinematics for up to 6-DOF serial robot arms. Includes PUMA/SCARA/UR presets. Real-time display of end-effector position and orientation.
Parameter Settings
Presets
DH Parameter Table
Joint
α[°]
a[mm]
d[mm]
θ[°]
Joint Angle Sliders
Results
—
X [mm]
—
Y [mm]
—
Z [mm]
—
Roll [deg]
—
Pitch [deg]
—
Yaw [deg]
—
Base-to-tip distance [mm]
—
DOF Count
Robot pose (orbit view)
Drag to orbit / double-click for the home view / pinch or Ctrl+wheel to zoom. Yellow dashed line = joint rotation axis zᵢ, cyan = tip trail
The bar chart above shows the absolute X and Z coordinates [mm] of each joint origin. The legend labels refer to joint position magnitude, not to the Jacobian ∂p/∂θ.
Theory & Key Formulas
The transformation matrix for each joint is constructed from DH parameters:
What exactly are these Denavit-Hartenberg parameters? They look like a bunch of angles and distances.
🎓
Basically, they're a standardized recipe for describing how robot links are connected. For each joint, you need just four numbers: a twist ($\alpha$), a link length ($a$), a joint offset ($d$), and a joint angle ($\theta$). In practice, this lets you mathematically "build" the robot's pose from the base to the end-effector. Try selecting the "PUMA" preset above—you'll see all six joints instantly populated with their classic DH parameters.
🙋
Wait, really? So if I change one of those joint angles ($\theta_i$) in the table, what happens?
🎓
You're telling that specific joint to rotate. The simulator then recalculates the entire chain of transformation matrices to find the new position and orientation of the robot's hand. For instance, try moving the slider for $\theta_2$ on the PUMA robot. You'll see the 3D model's elbow move, and the X/Y/Z and Roll/Pitch/Yaw values in the Results card will update in real time.
🙋
That final matrix is huge! How do I read it to know where the robot's hand actually is?
🎓
Great question. The tool reads it for you: the last column of $^0T_6$ gives the X, Y, Z coordinates of the end-effector in the base frame, and the 3x3 rotation block in the top-left is converted to roll–pitch–yaw (ZYX Euler angles) in the Results card. A common case is in welding robots—the coordinates tell you where the torch tip is, and the rotation tells you how it's angled against the car panel.
Physical Model & Key Equations
The core of forward kinematics is the homogeneous transformation matrix that moves from one joint frame to the next. It combines four basic transformations (two rotations and two translations) defined by the four DH parameters. This tool uses the modified DH convention (Craig), $^{i-1}T_i = R_x(\alpha_{i-1})\,T_x(a_{i-1})\,R_z(\theta_i)\,T_z(d_i)$, so the matrix is:
$\theta_i$ : Joint angle (rotation about Z-axis). $d_i$ : Link offset (translation along Z-axis). $a_{i-1}$ : Link length (translation along X-axis). $\alpha_{i-1}$: Link twist (rotation about X-axis).
The complete pose of the robot's end-effector relative to its base is found by chaining (multiplying) all the individual joint transformations together.
This single matrix $^0T_N$ contains everything: the 3D position is in the last column, and the 3x3 rotation matrix is in the top-left corner. This is the final output you see calculated by the simulator for all six joints.
Frequently Asked Questions
The table in this tool uses the column order α (twist), a (link length), d (offset), θ (joint angle). That is only the layout of the table, not a rule of the convention. Because the tool implements modified DH (Craig), α and a in each row belong to the previous link (i−1) while d and θ belong to joint i. Shifting columns or rows gives completely different poses, so confirm the definition of each parameter before input.
Yes, you can calculate any serial robot arm with up to 6 DOF by manually entering the DH parameters for each joint. The presets are provided as examples of typical parameters and can be edited freely (there is no save function). When entering a real robot's table, check whether the manufacturer uses standard DH or modified DH (this tool).
The position (x, y, z coordinates) and orientation (roll-pitch-yaw angles, ZYX Euler convention) are displayed in real time. The values update instantly, making it suitable for observing motion while changing θ values with sliders or similar controls.
This tool only supports forward kinematics. Inverse kinematics is difficult to implement generically because analytical solutions depend on the robot structure. However, you can use the forward kinematics results as a reference for verification when implementing numerical inverse kinematics separately.
Real-World Applications
Robotic Welding & Painting: In automotive assembly, robots follow precise paths. Engineers use forward kinematics to ensure the welding torch or paint nozzle reaches every point on a car door with the correct orientation. The DH parameters define the robot's geometry in the control software.
Surgical Robotics: For robotic arms assisting in surgery, calculating the exact tip position of a surgical tool from joint encoder readings is critical. Forward kinematics provides this in real-time, ensuring millimeter precision for procedures like laparoscopy.
CAE & Multibody Dynamics (MBD): When simulating a robot's movement in software like Adams or Simscape, each joint must be defined. The DH convention provides a clean, unambiguous way to set up these revolute and prismatic joints and their connecting links for dynamic analysis.
ROS & URDF Modeling: The Robot Operating System (ROS) uses URDF files to describe robots. These models are often derived from DH parameters. This simulator does not write URDF, but it lets you check the forward kinematics of a DH table before encoding the same geometry in URDF.
Common Misconceptions and Points to Note
When you start using DH parameters, it's easy to fall into a few common traps. A major misconception is thinking there is only one definition of DH parameters in the world. In reality, there are two main conventions: Standard DH and Modified DH. This tool uses Modified DH (Craig: $^{i-1}T_i = R_x(\alpha_{i-1})\,T_x(a_{i-1})\,R_z(\theta_i)\,T_z(d_i)$, the matrix shown under Theory & Key Formulas). Many textbooks and manufacturer datasheets use Standard DH, and entering such a table unchanged gives different results, so get into the habit of checking which convention is used first.
Next, pay attention to the sign and direction of parameters 'a' and 'α'. You might think the link length 'a' is always positive, but it can become negative depending on how you set up the coordinate system—for example, if you reverse the direction of a joint's axis of rotation. Also, the twist angle 'α' gets its sign from the right-hand rule: point your thumb along the X-axis, and the positive direction is from the Y-axis towards the Z-axis. Getting this wrong can cause the robot to move in a mirrored, opposite way.
Finally, a practical note: be cautious with calculation results at singular configurations. When a robot arm is fully extended or multiple joint axes align in a straight line—a "singular configuration"—the pose may not be uniquely determined mathematically. You can observe a related effect in the tool: with the PUMA preset, keep θ4=30° and move θ5 from 89° to 90° to 91°. The position does not change, but Roll jumps −180° → −90° → 0° and Yaw jumps −30° → 60° → 150°, because ZYX Euler angles are not unique when Pitch reaches ±90° (gimbal lock). This is a mathematical issue, not a bug in the tool, so keep it in mind.
Select a preset (PUMA, SCARA, UR, Stanford; UR is loaded at start) or enter your own DH table. Column order is α [°], a [mm], d [mm], θ [°]; the tool uses modified DH (Craig). All presets are laid out as modified-DH rows (αᵢ₋₁, aᵢ₋₁, dᵢ, θᵢ): PUMA uses a₂=432, a₃=20, d₃=149, d₄=433 mm (Craig's table rounded to integers) and UR uses the UR5 nominal values d₁=89, a₂=−425, a₃=−392, d₄=109, d₅=95, d₆=82 mm (the published standard-DH table shifted by one row)
Adjust joint angles with the sliders or number fields (−180° to +180°, degrees only). SCARA has 4 joints, the others 6
Every change recomputes the product of homogeneous transforms automatically (there is no run button) and updates X/Y/Z in mm, Roll/Pitch/Yaw in degrees (ZYX Euler) and the base-to-tip distance
The base-to-tip distance is for the current pose; it is not the reach of the robot
Worked Example
With the PUMA preset (a₂=432 mm, a₃=20 mm, d₃=149 mm, d₄=433 mm, Craig's modified-DH table rounded to integers) and all joint angles at 0°, the tool shows X=452.0 mm, Y=149.0 mm, Z=−433.0 mm and a base-to-tip distance of 643.4 mm. You can check this by hand: in modified DH a₂ and a₃ both lie along x, so X=432+20=452; d₃=149 appears in Y and d₄=433 in −Z; √(452²+149²+433²)=643.4. Setting θ₂=−90° alone gives X=433.0, Y=149.0, Z=452.0 (upper arm vertical) while the distance stays 643.4 mm. With the UR preset and all joints at 0° the tool shows X=−817.0, Y=−191.0, Z=−6.0 mm, which matches the UR5 nominal zero pose (−817.25, −191.45, −5.49) within rounding.
Practical Notes
The preset values are representative dimensions, not calibration data for a specific robot. Before entering a real robot's DH table, check whether it is written in standard DH or modified DH (this tool)
When Pitch approaches ±90°, Roll and Yaw jump discontinuously (gimbal lock of ZYX Euler angles). The orientation itself is continuous; only its representation is
In the SCARA preset all α are 0, so Z stays at 100 mm whatever θ₁ and θ₂ are; the tip moves only in the horizontal plane
Intermediate joint positions are drawn in the 3D view and plotted in the bar chart as |x| and |z| of each joint origin