Why not every degree of freedom needs its own motor — and why giving up one-motor-per-joint control is often the smarter design, not a shortcut.
Degrees of freedom (DOF) describe how many independent ways a mechanism's configuration can change — a purely kinematic count of joint parameters. Actuated jointsare a separate, much more practical number: how many of those joints actually have a motor, tendon, or other actuator driving them directly. It's easy to assume the two numbers should always match. They don't have to, and in a lot of good designs, they deliberately don't.
A rigid body free in 3D space has up to 6 degrees of freedom — three translations plus three rotations. For a mechanism built from connected links and joints, total DOF is the number of independent parameters needed to fully specify its configuration at any instant: fix that many joint angles (or positions), and the pose of every link in the mechanism is determined. DOF is a statement about kinematic freedom — how many ways the thing can move — not a statement about how many motors exist to move it.
Actuated joints count something else entirely: how many joints in the mechanism have their own dedicated motor or actuator driving them directly. A mechanism can have more degrees of freedom than actuated joints — this is called underactuation. The classic real-world example is an underactuated robotic gripper finger: a single motor or tendon can drive a finger with several knuckle joints, where the multiple joints (multiple DOF) passively conform to an object's shape as the finger closes, guided by mechanical linkages and springs rather than each knuckle having its own independent motor. The finger has more DOF (several knuckles that can each bend somewhat independently) than actuated joints (one motor or tendon driving the whole assembly).
That full independent control is genuinely useful — but it isn't free. Three motors means three sets of wiring, three drivers, three control-loop channels, and three times the mass and cost sitting out at the end of the arm, exactly where added weight hurts the most. For a lot of tasks, that price buys more precision than the task actually needs.
A fully actuated mechanism — one motor per degree of freedom — gives complete, independent control over every joint. That is real capability, but it is paid for in motors, wiring, drivers, control-loop bandwidth, added weight, and added cost, all multiplied by however many joints exist. An underactuated mechanism — fewer actuators than degrees of freedom — is deliberately simpler, lighter, and cheaper, and it can be functionally betterfor the task, not just cheaper. An underactuated gripper finger can passively self-adapt its shape to conform around an irregularly shaped object using a single motor, where a fully actuated finger would need per-joint sensing and a control algorithm to work out the same conforming shape on its own. The tradeoff is real: underactuated joints generally can't be independently, precisely commanded to arbitrary configurations. Their motion follows the passive mechanical or spring behavior designed into the linkage, not direct independent control — you get the adaptive behavior the linkage was designed to produce, and nothing else.
False, or at least badly incomplete. Underactuation — fewer actuators than degrees of freedom — is a deliberate, often advantageous design choice, not a compromise forced by budget alone. It reduces motor count, weight, cost, and control complexity, and in cases like adaptive gripper fingers, the passive mechanical compliance that underactuation provides can actually produce betterreal-world object-conforming behavior than a fully actuated finger would achieve without complex sensing and control specifically built to replicate it. The real tradeoff isn't "giving up capability" across the board — it's giving up independent, precise control of each individual joint in exchange for fewer motors, lower weight, lower cost, and, in the right applications, genuinely better passive behavior. Whether that trade is worth making depends entirely on whether the task needs precise per-joint positioning (a fully actuated design is the right call) or robust, low-cost conforming behavior around unknown shapes (underactuation is often the better engineering choice).
Explains the difference between degrees of freedom (DOF) — the number of independent ways a mechanism's configuration can change — and actuated joints, the number of joints that actually have their own dedicated motor. A mechanism can have more DOF than actuated joints, a design approach called underactuation, illustrated with a fully actuated three-knuckle robotic finger versus an underactuated single-tendon finger that passively conforms to an object's shape.
It's tempting to treat degrees of freedom and actuator count as the same number, since in the simplest possible mechanisms — a single revolute joint driven by a single motor — they happen to match. Once a mechanism has more than one joint, that assumption breaks down. DOF is a purely kinematic quantity: how many independent parameters describe the mechanism's configuration. It says nothing about how many of those parameters are directly driven by their own actuator. Confusing the two leads to the false conclusion that any mechanism with fewer motors than joints must be somehow incomplete or under-designed.
A mechanism is underactuated when it has more degrees of freedom than actuated joints. The clearest everyday example is an adaptive robotic gripper finger built with several knuckle joints (multiple DOF) but only a single motor or tendon driving the whole finger. As the tendon is pulled, it routes through a mechanical linkage across all the knuckles, and the joints bend in a sequence and amount determined by that linkage's passive mechanics — often assisted by return springs — rather than by independent commands to each knuckle. The result is a finger that can wrap around and conform to objects of different, even unknown, shapes using one actuator instead of three or more.
The alternative — a fully actuated finger with one motor per knuckle — can be commanded to any specific configuration on demand, precisely and independently at every joint. That capability costs real engineering resources: one motor, one driver, one wiring run, and one control channel per joint, all adding mass and cost concentrated at the end of the mechanism, where extra weight is most expensive to carry and move.
The right choice depends on what the task actually needs. Applications that require commanding each joint to an exact, known configuration — precision manipulation, repeatable pick-and-place at defined joint angles, force control at individual contact points — need full actuation, because underactuated joints simply cannot be driven to arbitrary configurations; their motion is set by the linkage's built-in passive behavior. Applications that need robust, low-cost adaptation to objects of varying or unknown shape — general-purpose grasping, prosthetic and low-cost robotic hands, compliant end effectors — are frequently better served by underactuation, since the passive compliance does real mechanical work (self-adapting contact) that would otherwise require additional sensors and a control algorithm to approximate, for a fraction of the actuator count, wiring, weight, and cost.
Not in a well-formed rigid mechanism — actuated joints are a subset of the joints contributing to DOF, so actuator count can be less than or equal to DOF, never more. If a mechanism appears to have redundant actuators exceeding its DOF, either some actuators are mechanically constrained together, or the mechanism is overconstrained/redundantly actuated, which is a distinct topic from underactuation.
No — DOF and actuation count are independent axes. A mechanism can have any number of DOF (1, 3, 6, or more) and separately be fully actuated (one motor per DOF) or underactuated (fewer motors than DOF) at that same DOF count. The finger example in this explainer has 3 DOF whether it's built with 3 actuators or 1.
No. The joints follow a designed, predictable mechanical relationship set by the linkage geometry and any springs — the motion isn't random, it's just not independently commandable to arbitrary angles. Pulling the single tendon a given amount produces a repeatable, designed pattern of knuckle motion; what's lost is the ability to set each knuckle's angle independently of the others.
You can approximate it, but it takes per-joint force or contact sensing plus a real-time control algorithm to replicate what an underactuated linkage does for free through its passive mechanics. That adds cost and complexity of its own (sensors, computation, tuning) — often more than the savings from fewer motors would have covered, which is exactly why underactuated designs remain the simpler, cheaper choice for object-conforming grasping.
Underactuated legged robots (fewer actuated joints than the leg's full kinematic DOF, relying on passive dynamics for some motion), tendon-driven prosthetic hands, and some compliant/soft robotic end effectors all use the same principle: fewer actuators than DOF, with the missing control authority replaced by designed passive mechanics.
Try our Mechatronics Studio
More calculators, simulators, and guides for this discipline.