Is pose the same as kinematic state?

For a rigid body, pose can describe its configuration. An articulated robot usually needs additional joint coordinates.

Does kinematic state always exclude velocity in every field?

No. Usage varies; this entry uses the AI robotics configuration convention and states that scope explicitly.

Can two joint configurations give the same end-effector position?

Yes. A position target can leave multiple arm arrangements possible, so the endpoint alone may not identify the full configuration.

AI Glossary / Graphs and symbolic AI / Intermediate

Kinematic state

Definition

A kinematic state, in the robotics convention used here, describes a robot's configuration: its position, orientation and any relevant joint positions. It specifies the robot's geometric arrangement at an instant. A dynamic state extends that description with quantities such as velocities needed to model how motion evolves.

What it means

For a rigid robot moving on a flat plane, a common configuration is q = (x, y, θ), where x and y locate a reference point and θ gives heading. A robot arm instead needs its joint coordinates, together with any relevant base configuration.

The state must contain enough geometric information for the intended model. Knowing only a gripper's position may leave the arm's elbow arrangement unknown. A fixed base can be defined as constant.

Terminology varies across engineering fields: some systems use kinematic state more broadly to include motion derivatives. This entry follows the configuration-versus-dynamic-state distinction used in the cited AI robotics material. Inspect the actual variables, not just the label.

How it works

Choose configuration coordinates and define their reference frames, units and conventions. Record or estimate their values at a stated time. A geometric model can then transform points from a robot-attached frame into the world frame, locate links and check configuration constraints.

Measurements do not automatically reveal the exact state. Joint encoders, cameras or localization systems can provide imperfect evidence, so an estimated configuration may also carry uncertainty. A belief about the state is different from the physical state itself.

A configuration describes where the robot is arranged, but it does not by itself say how quickly the robot is moving. Predicting motion under forces generally requires additional state and a dynamics model. A path through configuration space likewise needs timing and execution considerations before it becomes a complete motion description.

Three practical examples

A mobile-robot display shows a planar position and heading relative to a map.
A manipulator model uses joint positions to draw the current arrangement of its links.
A planning exercise tests whether a candidate configuration intersects an obstacle before considering how to reach it.

Think of it this way

Think of a labeled photograph of an articulated model, with every joint and reference axis identified. It can specify the arrangement, but a single photograph cannot tell whether the model is stationary or moving through that arrangement.

The analogy separates geometry from motion rate. A real configuration representation is numerical, may require more than one view to infer, and can be uncertain even when its display looks precise.

Worked example

The same location is not always the same configuration

Model a rigid robot on a plane. Its reference point is at world coordinates (2, 3) meters, and its heading θ is measured counterclockwise from the world's positive x direction. A marker lies one meter along the robot's forward axis, so its robot-frame coordinates are (1, 0).

At heading 0 degrees, the marker's world position is (3, 3). At heading 90 degrees, it is (2, 4). The reference point has not moved, but the configuration has changed because orientation is part of the state.

More generally, the marker coordinates are x_marker = 2 + cos θ and y_marker = 3 + sin θ, with the angle interpreted consistently. These equations illustrate why position alone cannot determine the location of every point on the body.

Now consider two robots with the same configuration q = (2, 3, 90 degrees). One is stationary. The other moves forward at a constant 1 meter per second with no turning. Under that simplified motion assumption, one second later their reference points are (2, 3) and (2, 4), respectively. Their initial kinematic states were equal, but their velocities differed.

The pose alone therefore cannot select between those futures. This representation exercise assumes ideal planar motion, exact coordinates and no obstacles.

Illustrative case study

The fictional Cedar Robotics team builds a simulator that stores only each robot's x and y position. Its rendered forward marker stays fixed even when the robot turns, because the representation has omitted heading.

The team adds heading, documents the reference frame and checks the two orientations in the worked example. Reviewers then notice that prediction still treats a moving and stationary robot identically. Cedar keeps configuration separate from velocity and expands the prediction model to use both.

The display distinguishes estimated poses from known test configurations. This fictional case shows why distinct situations need distinct state representations.

Common misconceptions and mistakes

Position alone always specifies the robot. Orientation and articulated joint positions can matter.
The same configuration implies the same future motion. Different velocities or controls can lead to different futures.
Six stored orientation numbers mean six rotational freedoms. Representation size and independent degrees of freedom are different concepts.
A precise-looking state display is exact ground truth. Its values may be estimates with uncertainty.

Business use

A clear configuration representation helps connect localization, planning, simulation and visualization. It also makes integration questions concrete: which frame does this position use, what does zero joint angle mean, and when was the measurement valid?

Specify these conventions in interfaces and test them using simple known configurations. For an articulated robot, check whether the state describes the whole mechanism or only an end-effector pose. When a project moves from geometric planning to execution, identify the additional velocity, actuator and dynamics information required. A collision-free configuration is useful evidence, but it is not a complete statement about a motion's feasibility.

Check your understanding

Why do the two robots at (2, 3, 90 degrees) have different predicted positions one second later?

Answer: Their configurations match, but their velocities do not. The stationary robot stays in place, while the other advances under the stated constant-velocity assumption.

Related terms

Learn next

Rotate the example robot to 180 degrees and calculate the marker position. Then add a joint coordinate for a movable arm and explain why the base pose no longer specifies the entire robot. Keep position, orientation, velocity and uncertainty in separate fields.

Pose

Sources

AIMA's robotics slides distinguish configuration-space planning from dynamic state including velocity: https://aima.cs.berkeley.edu/slides-pdf/chapter25.pdf

Lynch and Park explain rigid-body configuration freedoms: https://modernrobotics.northwestern.edu/nu-gm-book-resource/2-1-degrees-of-freedom-of-a-rigid-body/

Their treatment of constraints distinguishes configuration from permissible velocities: https://modernrobotics.northwestern.edu/nu-gm-book-resource/2-4-configuration-and-velocity-constraints/

The marker calculations and Cedar Robotics case are original illustrations.

Previous: Kimball lifecycle Next: Kinematics