Robot manipulator design, part 1: fundamentals

Degrees of freedom, joint types, common arm architectures and how to pick a configuration for a task.

A robot manipulator is a robot arm: a chain of links and joints that moves a tool to pick, place, weld, paint or assemble. You find them in car factories, operating theatres, warehouses and research labs. This first part covers the fundamentals you need before drawing a single part: degrees of freedom, joint types and the main arm layouts.

Four disciplines in one arm

Building an arm means getting four things to work together:

Area Covers
Mechanical design Links, joints, bearings, materials: strong but light
Electronics Motors, drivers, sensors, wiring, power
Control Making each joint reach its target precisely and safely
Programming Kinematics, motion planning and the task logic

Degrees of freedom

The degrees of freedom (DOF) of an arm are the number of independent joint motions. To place a tool at any position and any orientation in 3D space, an arm needs 6 DOF: three to position the wrist, three to orient the tool.

Fewer is often enough. A pick-and-place arm that always grips from above needs only 4 (x, y, z and a rotation about the vertical axis). More than 6 makes an arm redundant: it can reach the same pose in many ways and use the extra freedom to avoid obstacles, like a human arm with 7.

Joint types

Joint Symbol Motion Example
Revolute R Rotation about an axis Shoulder and elbow of most arms
Prismatic P Sliding along an axis Vertical axis of a SCARA, gantry axes
Spherical S Rotation about three axes (ball joint) Wrists, passive joints in parallel robots

An arm is often named by its first three joints, from the base outwards: an RRP arm has two revolute joints followed by a prismatic one.

The common arm layouts

Layout Joints Workspace Strengths Typical use
Cartesian (gantry) PPP Box Simple control, very accurate, scales to large sizes 3D printers, CNC, palletising
Cylindrical RPP Cylinder Simple, good vertical reach Machine loading
Spherical (polar) RRP Part of a sphere Long reach Early industrial robots
SCARA RRP (+ wrist R) Flat cylinder Very fast in a plane, stiff vertically Assembly, pick and place
Articulated RRR + 3R wrist Nearly a sphere Flexible, reaches around objects Welding, painting, general use
Parallel (delta) Several chains to one platform Small dome Extremely fast and light, stiff Food packing, sorting

Picking a configuration

Start from the task, not from the robot:

$$ \text{Task} \;\rightarrow\; \text{Required DOF} \;\rightarrow\; \text{Workspace shape} \;\rightarrow\; \text{Layout} \;\rightarrow\; \text{Link lengths and motors} $$

  • Moving parts between points on a table, quickly? → SCARA or delta.
  • Very large area, high accuracy, simple control? → Cartesian.
  • Reaching around obstacles or holding a tool at any angle? → articulated, 6 DOF.
  • Heavy loads at low speed? → short links, large gear reduction, possibly Cartesian.

A useful rule: use the fewest joints that do the job. Every extra joint adds weight, cost, error and control effort.

How much can the arm lift?

Every joint has to carry everything beyond it, so the shoulder works hardest. A first estimate of the holding torque at a joint, with the arm stretched out horizontally, is

$$ \tau \approx g \left( m_\text{load}\, d_\text{load} + \sum_i m_i\, d_i \right) $$

where each mass \(m_i\) (links, motors further out) acts at horizontal distance \(d_i\) from the joint. This is why designers put motors close to the base and make the outer links as light as possible.

Our experience

When we built our own manipulator, the biggest lesson was how closely the four areas depend on each other: a heavier motor changes the link design, which changes the torque, which changes the motor again. Working through mechanical design, electronics, control and programming together, rather than one after the other, is what finally gave us an arm that performed its tasks reliably.

Coming in part 2

Part 2 looks at workspace analysis, singularities, redundancy, how to measure an arm's performance and how to control it.

Watch the video

For the full explanation, see Robot manipulator design on YouTube.

Basheer Al-Tawil is a PhD researcher in Robotics and AI at Otto von Guericke University. Found a mistake or have a question? Write to me.

Browse by topic