Actuation, End Effectors, and Robot Control

Introduction and Context

A robot becomes useful only when it can generate controlled motion and interact safely with its environment. The mechanical structure determines the robot’s possible motions, but actuators supply energy, transmissions convert that energy into useful torque or force, end effectors perform task-specific work, and controllers coordinate all components. These elements form an integrated mechatronic system rather than independent subsystems.

In industrial manipulators, actuator selection affects speed, payload, accuracy, noise, energy use, maintenance, and safety. A gripper may determine whether a product is damaged or successfully transferred. The control strategy determines whether the robot merely follows a trajectory or can also regulate contact force, compensate for disturbances, and adapt to uncertainty.

This chapter compares major actuator technologies, examines drives and transmissions, classifies end effectors, and develops the principal control concepts used in robotic systems.

Learning Objectives

After studying this chapter, the reader should be able to:

1. Compare electric, hydraulic, pneumatic, compliant, and soft actuators.
2. Select a motor, drive, and transmission for a robotic joint.
3. Match grippers and tools to task, object, and environmental requirements.
4. Explain joint, position, velocity, force, impedance, hybrid, and model-based control.
5. Apply fundamental actuator and robot-control equations.
6. Evaluate practical, safety, ethical, and maintenance implications of an actuation and control design.

1. Actuation as an Energy-Conversion Problem

An actuator converts an input energy form into mechanical motion. The actuator itself is only one part of the power chain:

energy source → power electronics or valve → actuator → transmission → joint or tool → environment

The principal performance variables are:

• Torque or force: mechanical output capability.
• Speed: angular speed for rotary joints or linear speed for prismatic joints.
• Power: rate of mechanical energy transfer.
• Bandwidth: how rapidly output can respond to a changing command.
• Backdrivability: ease with which an external load drives the actuator backward.
• Efficiency: useful mechanical power divided by supplied power.
• Stiffness and compliance: resistance, or controlled yielding, under displacement.
• Duty cycle: proportion of operating time spent delivering load.
• Controllability: ability to regulate position, velocity, torque, or force.

For a rotary actuator, mechanical power is

where P is power in watts, τ is torque in newton-metres, and ω is angular velocity in radians per second. For a translating actuator,

where F is force in newtons and v is linear velocity in metres per second. These equations describe instantaneous mechanical power and assume consistent SI units.

A practical actuator must satisfy both peak and continuous requirements. Peak torque may be needed during acceleration, while continuous torque is limited by thermal capacity. Selecting only for nominal load can cause overheating or loss of motion during transients.

2. Actuator Technologies

2.1 Electric Actuators

Electric actuation is the dominant technology in many industrial and collaborative robots. Common motors include brushed and brushless DC motors, permanent-magnet synchronous motors, and induction motors. Brushless and synchronous motors are especially common because they provide high power density, long service life, and precise electronic commutation.

Advantages include high efficiency, clean operation, precise feedback, straightforward integration with digital drives, and regenerative braking. Disadvantages include limited overload tolerance compared with some hydraulic systems, possible gearbox backlash, and thermal limitations at low speed or high holding torque.

Motor torque is related to current approximately by

where τ_m is motor torque, k_t is the torque constant in N·m/A, and i is motor current in amperes. This relation is an idealized steady-state model; magnetic saturation, friction, temperature, and drive limits can produce deviations.

2.2 Hydraulic Actuators

Hydraulic cylinders and motors use pressurized fluid. Their major strengths are high force density, overload capability, and suitability for large manipulators, mobile robots, and heavy construction equipment. Hydraulic systems can produce very large forces from compact actuators.

For an ideal hydraulic cylinder,

where F is cylinder force, p is pressure difference in pascals, and A is effective piston area in square metres. Real output is reduced by seal friction, leakage, pressure losses, and valve dynamics.

Hydraulics require pumps, reservoirs, filters, valves, hoses, and maintenance. Leakage creates environmental and cleanliness concerns. Compressibility of fluid and hose flexibility also complicate high-bandwidth control. Hydraulic actuation is appropriate when force density is more important than cleanliness, simplicity, or energy efficiency.

2.3 Pneumatic Actuators

Pneumatic cylinders and rotary vane actuators use compressed air. They are inexpensive, lightweight, fast, and intrinsically compliant because air is compressible. Pneumatics are widely used for simple pick-and-place operations, clamping, and process tools.

Their disadvantages are poor positioning precision, variable force with pressure and volume, compressor noise, and low energy efficiency. Proportional valves and servo-pneumatic systems improve control but increase cost and complexity. Pneumatics are often selected when the task needs rapid binary motion rather than accurate continuous positioning.

2.4 Compliant Actuators

A compliant actuator deliberately permits relative motion or force-dependent deformation between the motor and load. Compliance may be passive, using springs and elastic elements, or active, using torque sensing and control. Series elastic actuators place an elastic element between motor and load; its deformation estimates transmitted torque.

For a linear elastic element,

where F is elastic force, k is stiffness in N/m, and x is deflection in metres. This is Hooke’s law and assumes approximately linear elastic behaviour within the element’s operating range.

Compliance improves impact tolerance, force measurement, and human safety, but reduces positional rigidity and may introduce resonances. It is valuable in legged robots, collaborative manipulators, rehabilitation devices, and tasks involving uncertain contact.

2.5 Soft Actuators

Soft actuators are constructed from flexible materials and often use pneumatic networks, tendons, shape-memory alloys, electroactive polymers, or fluidic chambers. Their deformation is distributed rather than concentrated at rigid joints. Soft systems can conform to irregular objects and tolerate collisions.

Their principal challenges are nonlinear deformation, hysteresis, limited modelling accuracy, low repeatability, difficult sensing, and sometimes slow response. They are especially promising for delicate grasping, wearable robots, biomedical devices, and manipulation in unstructured environments.

Comparative Selection

No technology is universally superior. Selection should begin with task requirements, environment, payload, required precision, available utilities, safety constraints, and life-cycle cost.

3. Drives and Transmissions

A drive regulates energy delivered to an actuator. An electric servo drive usually contains a power converter, current controller, position or velocity loops, feedback interface, and protection functions. Hydraulic and pneumatic drives use pumps, valves, pressure regulation, and flow control.

A transmission adapts actuator speed, torque, force, or motion geometry to the joint. Common transmissions include spur and planetary gears, harmonic or strain-wave drives, cycloidal reducers, belts, chains, lead screws, ball screws, cable drives, and direct-drive arrangements.

For an ideal gear reduction with ratio N = ω_m/ω_j,

and

where ω_m and ω_j are motor and joint angular velocities, τ_m and τ_j are motor and joint torques, N is greater than one for reduction, and η is transmission efficiency between 0 and 1. These equations assume a fixed ratio and neglect backlash, elasticity, and speed-dependent losses.

A high reduction ratio increases output torque but lowers speed, reduces backdrivability, and can amplify reflected inertia and friction. Direct-drive motors eliminate gearbox backlash and can be highly backdrivable, but they require large motors and high current. Harmonic drives offer high reduction and compactness with low backlash, but their flexsplines have fatigue limits. Belts and cables are light and quiet but require tension management and may stretch.

Worked Example: Motor and Gearbox Selection

A joint must provide 30 N·m continuously and 45 N·m for short acceleration intervals. A motor can supply 2 N·m continuously and 3 N·m peak. A gearbox has ratio N = 20 and efficiency η = 0.85.

Continuous output torque is

τ_j = ηNτ_m = 0.85 × 20 × 2 = 34 N·m

Peak output torque is

τ_j,peak = 0.85 × 20 × 3 = 51 N·m

Thus the selection meets both torque requirements, assuming thermal, speed, duty-cycle, and structural limits are also satisfied. It would be incorrect to conclude suitability from torque alone: the gearbox must also withstand the load, and the motor must operate below its thermal limit.

4. End Effectors and Tool Selection

An end effector is the device attached to the robot flange that performs the task. It may grasp an object, process a surface, sense a condition, or apply a manufacturing operation.

4.1 Grippers

Parallel-jaw grippers provide simple, robust motion and are suitable for regular objects. Angular grippers rotate jaws away from the object and can provide clearance. Three-finger grippers centre approximately cylindrical or irregular parts. Adaptive and underactuated grippers use compliance to conform to object geometry. Vacuum grippers are effective for smooth, nonporous surfaces but require adequate seal area. Magnetic grippers handle ferromagnetic objects but cannot grasp nonmagnetic materials and may retain unwanted particles. Electroadhesive and gecko-inspired grippers can handle special surfaces but may be sensitive to contamination.

Gripper selection should consider object mass, centre of gravity, surface roughness, temperature, fragility, acceleration, orientation, contamination, cycle time, and failure consequences.

For a frictional grasp with two opposing contact surfaces, a simplified static condition is

where μ is the coefficient of friction, N is normal force at each contact, m is object mass, g is gravitational acceleration, a is additional downward acceleration, and S is a safety factor. This model assumes two symmetric contacts, no external moments, and load acting in the friction direction. Real designs must also check tipping moments, jaw geometry, contact pressure, and actuator failure.

4.2 Process Tools

Tools include welding torches, deburring spindles, screwdrivers, drills, dispensing nozzles, cutters, polishing heads, and inspection sensors. A process tool should be selected not only for its task output but also for reaction forces, heat, vibration, cable routing, calibration, maintenance, and guarding.

Tool changers improve flexibility but introduce mass, cost, alignment error, and potential connection failure. A tool interface should provide mechanical locking, utility connections, repeatable locating, presence detection, and safe confirmation before operation.

5. Robot Control Architecture

Robot control is commonly organized in nested loops:

1. Current or torque loop: regulates motor current and therefore approximate motor torque.
2. Velocity loop: regulates joint speed.
3. Position loop: regulates joint angle or Cartesian position.
4. Task and supervisory layers: generate trajectories, select tools, manage modes, and monitor safety.

Inner loops are normally faster than outer loops. This separation simplifies design, but interactions among sampling, computation, actuator dynamics, friction, structural flexibility, and communication delays must be considered.

5.1 Joint-Space Position Control

For joint coordinate q, a proportional-derivative controller is

where q_d and ẋ_d are desired position and velocity, q and ẋ are measured position and velocity, and K_p and K_d are proportional and derivative gains. In standard notation, ẋ here denotes joint velocity; many robotics texts use ḋq for the same quantity.

For a constant desired position, the velocity term is usually zero. PD control gives responsive correction but may exhibit steady error under gravity or persistent loads. Integral action can remove constant error:

where e = q_d − q, and K_i is the integral gain. Integral action must be limited or anti-windup methods used when torque saturates.

5.2 Velocity Control

Velocity control regulates joint speed, often for mobile bases, continuous rotation, or process operations. A simple controller uses velocity error, but accurate performance requires compensation for friction, load torque, and acceleration demands. Velocity commands should be bounded to prevent excessive kinetic energy.

5.3 Force and Torque Control

Force control regulates interaction force rather than only position. A force sensor may be mounted at the wrist, or joint torque may be estimated from motor current and a transmission model. Force control is essential for polishing, insertion, assembly, grasping, and human–robot interaction.

A basic force controller can be written as

where ẋ_c is commanded Cartesian velocity, F_d is desired force, F is measured force, and K_f is a force-control gain. This expression is a conceptual admittance law; stability depends on sampling, contact stiffness, delay, and the surrounding position loop.

5.4 Impedance and Admittance Control

Impedance control specifies a dynamic relationship between position error and interaction force. A desired mass–spring–damper behaviour is

where M_d, B_d, and K_d are desired inertia, damping, and stiffness matrices; x is Cartesian position; and F is external force. The equation assumes a locally defined Cartesian frame and appropriately dimensioned matrices.

An impedance controller is particularly useful when the robot should yield under contact while still returning toward a nominal path. Admittance control instead measures force and generates a motion command through a virtual mass–damper–spring model. High-stiffness robots with reliable force sensing often use admittance control, whereas backdrivable robots can implement impedance more directly.

5.5 Hybrid Position–Force Control

Hybrid control assigns position regulation to selected task-space directions and force regulation to complementary directions. For example, during surface polishing, motion tangential to the surface may be position-controlled while normal force is regulated. A selection matrix identifies the controlled subspaces. The position and force directions must be physically compatible; attempting to independently command both position and force in the same unconstrained direction generally creates conflict.

5.6 Model-Based Control

The rigid-body dynamics of an n-joint robot are commonly expressed as

where M(q) is the inertia matrix, C(q,q̇)q̇ represents Coriolis and centrifugal effects, g(q) is gravity torque, τ_f is friction torque, τ_ext is external torque, and τ is commanded joint torque. The model assumes rigid links and a chosen coordinate convention; flexibility, actuator dynamics, backlash, and unmodelled loads may require additional terms.

Computed-torque control uses the model to compensate for known dynamics and applies feedback to the remaining tracking error. Its advantage is improved performance over a broad operating range. Its limitation is sensitivity to parameter uncertainty, friction, payload changes, and inaccurate contact models. In practice, model-based control is usually combined with feedback rather than used alone.

6. Worked Control Example

Consider a one-degree-of-freedom joint with desired angle 0.50 rad and measured angle 0.46 rad. The desired and measured velocities are both zero. Let K_p = 40 N·m/rad and K_d = 2 N·m·s/rad. The position error is

e = 0.50 − 0.46 = 0.04 rad

Therefore,

τ_c = 40(0.04) + 2(0) = 1.6 N·m

The command is 1.6 N·m before gravity compensation, friction compensation, saturation, and safety limits. If a gravity-compensation term of 0.8 N·m is required in the same direction, the total command becomes 2.4 N·m, provided the actuator can safely deliver it.

7. Applied Case: Collaborative Assembly of a Fragile Connector

Background

Consider an illustrative automated workstation that inserts a delicate electrical connector into a housing. The connector is light but easily damaged by excessive lateral force. Its position varies slightly because parts arrive in trays with manufacturing tolerances.

Problem

A rigid position-only robot occasionally jams or bends the connector. Increasing positional gain improves repeatability but increases impact force. A vacuum gripper is unsuitable because the connector has a porous and irregular surface.

Analysis

The task requires accurate approach motion, compliant contact, force limitation, and detection of successful insertion. A parallel-jaw gripper with soft, replaceable pads provides stable grasping. An electric servo joint with a low-backlash gearbox supplies adequate precision, while a wrist force–torque sensor detects contact. The controller should not use pure position control during insertion because position error can represent either misalignment or contact deformation.

Approach and Solution

The robot first uses joint-space or Cartesian position control to approach the housing. At the insertion surface, a hybrid strategy regulates lateral position and axial force. Impedance or admittance behaviour allows small lateral corrections while limiting contact force. A force threshold and insertion-depth condition are used for state transitions. The controller includes velocity limits, torque limits, collision monitoring, and a retreat routine if force rises unexpectedly.

Outcome

The resulting system is more tolerant of tray variation and reduces connector damage. Its performance depends on sensor calibration, correct force-frame definition, suitable compliance, and carefully tuned transition logic. The solution also requires guarding and validated fault responses because a force sensor alone cannot guarantee safety.

Learning Points

1. Actuator, gripper, sensor, and controller choices must be made together.
2. Compliance can improve task success but does not replace sensing or safety design.
3. Hybrid control is appropriate when different task directions have different objectives.
4. A robust industrial system includes fault detection, recovery, and maintenance procedures.

8. Safety, Ethics, and Professional Practice

Actuation creates hazards through crushing, impact, cutting, entanglement, stored pneumatic or hydraulic energy, and unexpected restart. A risk assessment should identify hazards, estimate risk, implement inherently safe design where possible, and verify protective measures. Emergency stops, safe torque off, speed and separation monitoring, mechanical stops, pressure relief, guarding, and validated safety-rated control may all be necessary.

Collaborative operation must not be assumed merely because a robot is marketed as collaborative. The complete application—including tool, payload, workpiece, speed, posture, and foreseeable misuse—determines risk. Force limits should be validated experimentally and against applicable safety standards and local regulations.

Professional engineers should document assumptions, calibration procedures, software versions, actuator limits, maintenance intervals, and failure modes. Data from force sensors and cameras may reveal worker behaviour or production information; collection and access should therefore follow privacy and cybersecurity policies. Machine-learning controllers require additional validation because performance can change with data distribution, lighting, wear, or unexpected objects. Safety-critical functions should remain deterministic, monitored, and independently verified.

9. Contemporary Developments and Best Practices

Current research and industry development emphasise direct-drive and quasi-direct-drive actuators, integrated torque sensing, variable-stiffness mechanisms, soft grippers, energy regeneration, and digital twins. Learning-based methods can assist grasp selection, calibration, and model adaptation, but they should complement—not obscure—physical safety limits.

Best practices include:

• Define task forces, speeds, precision, duty cycle, and environment before choosing hardware.
• Size actuators for continuous and peak loads, including acceleration and payload uncertainty.
• Evaluate transmission backlash, compliance, efficiency, reflected inertia, and service life.
• Select end-effectors using object geometry, friction, fragility, contamination, and failure consequences.
• Use calibrated sensing and define coordinate frames explicitly.
• Tune nested control loops from inner to outer loops and check sampling and delay.
• Add saturation handling, anti-windup, fault detection, and safe recovery.
• Validate contact control with realistic variation in payload, friction, and alignment.
• Record maintenance, calibration, and software configuration for traceability.

Chapter Summary

Actuators convert energy into controlled mechanical action, while drives and transmissions adapt that action to robot joints and tools. Electric actuators offer precise and clean control; hydraulic systems provide exceptional force density; pneumatics provide fast, simple motion; compliant and soft actuators improve contact tolerance and conformability but introduce modelling and control challenges.

Transmission ratio, efficiency, backlash, elasticity, speed, and backdrivability are central to joint selection. End effectors must be matched to object properties, task forces, environmental conditions, and safety requirements. Robot control ranges from joint position and velocity regulation to force, impedance, admittance, hybrid, and model-based control. Position control is effective for free-space motion, whereas force-related strategies are essential for contact tasks. Model-based terms improve performance but must be supported by feedback because real robots contain friction, compliance, uncertainty, and changing payloads.

A successful robotic application integrates actuator, transmission, end effector, sensing, control, safety, and maintenance decisions from the beginning.

Key Terms

Actuator: Device that converts energy into mechanical motion.

Backdrivability: Ease with which an external load drives an actuator or joint.

Bandwidth: Frequency range over which a system responds adequately to commands.

Compliance: Ability to deform or yield under applied force.

Computed-torque control: Model-based control that compensates for robot dynamics.

Drive: Power-regulation subsystem supplying controlled energy to an actuator.

End effector: Task-specific device mounted at the robot flange.

Force control: Regulation of interaction force rather than only position.

Gear ratio: Ratio relating motor speed to joint speed in a transmission.

Gripper: End effector designed to hold or manipulate objects.

Hybrid control: Simultaneous position control and force control in complementary directions.

Impedance control: Control of the dynamic relationship between motion and force.

Payload: Mass and associated inertial load carried by a robot.

Pneumatic actuator: Actuator powered by compressed gas.

Soft actuator: Flexible actuator producing distributed deformation.

Torque control: Regulation of rotational effort at a joint.

Transmission: Mechanism that transfers and transforms actuator motion and force.

Variable stiffness: Capability to change mechanical or controlled compliance.

Velocity control: Regulation of joint or tool speed.

Review Questions

Short Answer Questions

1. Why must both continuous and peak actuator torque be considered in robot design?
2. State two advantages and two limitations of hydraulic actuation.
3. What is the purpose of a transmission in a robotic joint?
4. Distinguish impedance control from force control.
5. Why is compliance useful during contact-rich tasks?

Long Answer Questions

1. Compare electric, hydraulic, pneumatic, compliant, and soft actuators for industrial robotics.
2. Explain how gearbox ratio, efficiency, backlash, and backdrivability influence joint performance.
3. Describe the selection process for a gripper handling fragile, irregular objects.
4. Explain the structure and purpose of nested current, velocity, and position control loops.
5. Discuss the assumptions and limitations of model-based robot control.

Analytical / Application Questions

1. A motor produces 2.5 N·m and drives a joint through a 15:1 reducer with 80% efficiency. Calculate the idealized joint torque and explain two additional checks required before approving the design.
2. Design a control strategy for robotic surface polishing in which normal force must remain constant while tangential motion follows a programmed path. Identify the controlled variables, sensors, and likely failure modes.
3. Select an actuator and end effector for a robot that transfers hot, porous ceramic parts. Justify the choices using force, temperature, contamination, compliance, safety, and maintenance criteria.

Multiple Choice Questions with Answers

Question 1

Which actuator technology generally provides the highest force density for heavy robotic machinery?

A. Hydraulic

B. Pneumatic

C. Small brushed electric

D. Soft elastomeric

Correct Answer: A. Hydraulic

Question 2

For an ideal gear reduction with ratio N greater than 1, which relationship is correct when efficiency is η?

A. ω_j = Nω_m and τ_j = τ_m/N

B. ω_j = ω_m/N and τ_j = ηNτ_m

C. ω_j = ηNω_m and τ_j = τ_m

D. ω_j = ω_m and τ_j = ητ_m/N

Correct Answer: B. ω_j = ω_m/N and τ_j = ηNτ_m

Question 3

A motor supplies 2 N·m to a 10:1 gearbox with 90% efficiency. What is the approximate output torque?

A. 1.8 N·m

B. 18 N·m

C. 22.2 N·m

D. 20 N·m

Correct Answer: B. 18 N·m

Question 4

Which end effector is usually most dependent on a good seal and a nonporous object surface?

A. Magnetic gripper

B. Three-finger gripper

C. Vacuum gripper

D. Parallel-jaw mechanical gripper

Correct Answer: C. Vacuum gripper

Question 5

What is the principal purpose of derivative action in a PD joint controller?

A. To increase steady-state error

B. To respond to velocity error and provide damping

C. To eliminate all actuator saturation

D. To measure payload mass directly

Correct Answer: B. To respond to velocity error and provide damping

Question 6

Which control approach is most appropriate for regulating normal polishing force while controlling tangential tool motion by position?

A. Pure joint-position control

B. Hybrid position–force control

C. Open-loop velocity control

D. Binary pneumatic control

Correct Answer: B. Hybrid position–force control

Question 7

In the simplified friction-grasp relation 2μN ≥ S m(g + a), what does μ represent?

A. Motor efficiency

B. Gear ratio

C. Coefficient of friction

D. Joint inertia

Correct Answer: C. Coefficient of friction

Question 8

What is a major benefit of a series elastic actuator?

A. It eliminates all resonances

B. It provides compliance and can support torque estimation from deflection

C. It requires no feedback

D. It always increases positional rigidity

Correct Answer: B. It provides compliance and can support torque estimation from deflection

Question 9

Which term in the robot dynamic equation represents gravity torque?

A. M(q)q̈

B. C(q,q̇)q̇

C. g(q)

D. τ_f(q̇)

Correct Answer: C. g(q)

Question 10

Why is model-based control normally combined with feedback?

A. Robot dynamics are perfectly known only at zero speed

B. Feedback is needed to address modelling errors, friction, disturbances, and changing loads

C. Feedback removes the need for actuators

D. Model-based control cannot use torque commands

Correct Answer: B. Feedback is needed to address modelling errors, friction, disturbances, and changing loads