Industrial Robotics and Human–Machine Collaboration

Introduction and Context

Industrial robotics combines mechanical systems, sensing, control, programming, and production engineering to perform repeatable tasks in manufacturing. Robots are used for material handling, machine tending, assembly, welding, painting, inspection, packaging, and hazardous operations. Their value is not simply that they replace manual labour. Properly designed robotic systems improve consistency, ergonomics, traceability, throughput, and process capability while allowing people to perform higher-value activities such as supervision, maintenance, problem solving, and process improvement.

Modern manufacturing increasingly combines conventional industrial robots with collaborative robots, machine vision, force sensing, digital simulation, and data analytics. The central engineering problem is therefore socio-technical: the robot, tooling, software, workplace, product, and human operators must be designed as one system.

Learning Objectives

After studying this chapter, students should be able to:

1. Classify common robot configurations and select a suitable configuration for a task.
2. Explain robot kinematics, workspace, repeatability, accuracy, and payload.
3. Select end effectors, sensors, and programming methods for manufacturing operations.
4. Design a safe and efficient robotic work cell.
5. Compare conventional guarded robots with collaborative applications.
6. Evaluate robot performance, economic feasibility, and implementation risks.
7. Analyse a practical robotic manufacturing application.

1. Robot Architecture and Configurations

An industrial robot is a programmable manipulator consisting of a mechanical structure, actuators, joints, controllers, power systems, software, and an end effector. A robot’s degrees of freedom (DOF) are the independent motions required to define its configuration. A typical six-axis articulated robot has three axes to position the wrist and three axes to orient the tool.

1.1 Joint types and configurations

A revolute joint produces rotation, while a prismatic joint produces linear translation. Robot configurations are commonly classified as follows:

Configuration selection depends on workspace shape, reach, payload, speed, stiffness, dexterity, access, floor area, and cost. A SCARA robot is efficient for planar assembly but less suitable for complex three-dimensional orientation. An articulated robot offers broad orientation capability but generally has more complicated kinematics and collision risks.

1.2 Workspace and singularity

The workspace is the set of positions and orientations the robot can reach. The reachable workspace includes all attainable points; the dexterous workspace includes points attainable with the required range of orientations. Obstacles, joint limits, cables, tooling, and safety boundaries reduce the usable workspace.

A singularity is a configuration in which the robot loses motion capability in one or more Cartesian directions or requires very high joint velocities to produce a small Cartesian motion. Near a singularity, motion may become unpredictable, vibration may increase, and controller speed may be limited. Cell designers should avoid placing critical paths near singular configurations and should verify trajectories through simulation and commissioning tests.

2. Robot Kinematics and Performance

Kinematics describes motion without considering forces. Forward kinematics determines tool position and orientation from joint variables. Inverse kinematics determines joint variables needed to achieve a desired tool pose. A robot pose includes position and orientation, commonly represented by a homogeneous transformation matrix:

Here, R is a 3 × 3 rotation matrix, p is a 3 × 1 position vector, and the bottom row is [0 0 0 1]. The matrix is dimensionless in its rotational part, while the position components have units of length. Homogeneous transformations are useful because successive coordinate-frame transformations can be multiplied:

This formulation assumes rigid links and coordinate frames defined consistently. The multiplication order is essential because matrix multiplication is not generally commutative.

For a planar two-link robot, with link lengths L₁ and L₂ and joint angles θ₁ and θ₂, the end-point coordinates are:

Lengths must use the same unit, and angles must be interpreted consistently. These equations illustrate forward kinematics; solving them for the angles is an inverse-kinematics problem that may have multiple solutions or no solution within joint limits.

The Jacobian relates joint velocities to Cartesian velocity:

where v is the Cartesian velocity vector, J(q) is the configuration-dependent Jacobian, q̇ is the vector of joint velocities, and q is the vector of joint positions. The Jacobian is central to speed control, force control, singularity analysis, and robot calibration.

Accuracy and repeatability

Accuracy is the difference between the commanded or nominal position and the actual position. Repeatability is the closeness of repeated positions under the same conditions. A robot can have excellent repeatability but poor absolute accuracy if its geometric model, tool centre point, or base frame is incorrectly calibrated. High repeatability is often sufficient for machine tending when fixtures locate parts consistently; high absolute accuracy is more important in offline programming, precision assembly, and large-part processing.

Sources of error include link-geometry errors, joint backlash, compliance, thermal expansion, controller interpolation, payload variation, calibration errors, and fixture movement. Accuracy can be improved through mechanical design, tool and frame calibration, external measurement systems, thermal compensation, and error mapping.

3. End Effectors, Sensors, and Process Equipment

The end effector connects the robot to the process. Its design must consider payload, centre of gravity, gripping force, part variation, surface condition, cycle time, compliance, maintenance, and failure response.

3.1 End-effector types

• Mechanical grippers: parallel, three-finger, angular, or custom jaws; suitable for rigid parts.
• Vacuum grippers: effective for smooth, nonporous surfaces; require monitoring for pressure loss.
• Magnetic grippers: useful for ferromagnetic components but unsuitable where residual magnetism is unacceptable.
• Adhesive and electrostatic grippers: useful for special materials but sensitive to contamination and environment.
• Process tools: welding guns, deburring spindles, dispensing nozzles, screwdrivers, drills, and inspection probes.

A gripper must retain the workpiece under acceleration. For a simple friction grip, the available friction force is approximately:

where F_f is friction force in newtons, μ is the coefficient of friction, and N is the normal gripping force in newtons. If two opposing jaws support a vertically accelerated part, a conservative design requires:

where S is a safety factor, m is mass in kilograms, g is gravitational acceleration in m/s², and a is upward acceleration in m/s². This simplified model assumes equal load sharing, friction as the primary retention mechanism, and no mechanical form-fit. Real designs must also consider torque caused by the centre-of-gravity offset, vibration, leakage, jaw wear, and emergency stopping.

Worked example: gripper force

A 4 kg component is lifted vertically with an acceleration of 2 m/s². Two jaws share the load, the coefficient of friction is 0.30, and the design safety factor is 2. Using g = 9.81 m/s²:

Required total friction force:

F = 2 × 4 × (9.81 + 2) = 94.48 N

Thus, the required total normal force is N_total = 94.48 / 0.30 = 314.9 N. If the jaws share the force equally, each jaw must provide approximately 157.5 N. The actuator should be selected above this theoretical value to allow for wear, contamination, pressure variation, and dynamic effects.

3.2 Sensors

Sensors may be internal or external. Encoders measure joint position; torque sensors estimate joint effort; proximity sensors detect parts; machine vision identifies position, orientation, defects, and presence; force–torque sensors measure interaction forces; and laser or structured-light systems measure geometry.

Sensor selection should begin with the decision requirement. A presence sensor may be adequate for a fixed fixture, whereas vision is required when part location varies. Force sensing is useful for insertion, polishing, and delicate contact, but it increases integration complexity and requires filtering and carefully defined force limits.

4. Programming and Control

Robot programming methods include teach pendant programming, lead-through programming, offline programming, and programming through higher-level manufacturing systems. Teach pendant programming is accessible for small changes but can interrupt production. Offline programming supports cell layout, reachability analysis, cycle-time estimation, and collision checking, although the virtual model must be calibrated against the physical cell.

A robust robot program separates:

1. Motion logic: joint, linear, or circular movement.
2. Process logic: welding, gripping, dispensing, or inspection commands.
3. I/O and interlocks: confirmation of clamps, doors, part presence, and machine status.
4. Error handling: safe stopping, alarms, recovery, and operator guidance.
5. Data management: recipes, traceability records, and quality results.

Joint motion is often efficient for moving between distant points, while linear motion is preferred where the tool must follow a controlled path. Speed, acceleration, blending, and approach distances influence both cycle time and process quality. Programs should use safe approach and retreat positions, avoid unnecessary travel, and verify that the tool centre point is correctly defined.

Robot control may be position-based, velocity-based, force-based, or impedance-based. In impedance control, the robot is programmed to behave like a virtual mass–spring–damper system, allowing controlled compliance during contact. Such control is valuable for insertion and surface finishing but requires suitable sensing, tuning, and stability verification.

5. Robotic Cell Design

A robotic cell includes the robot, tooling, fixtures, conveyors, machines, sensors, controllers, guarding, operator interfaces, and material flow. Design should begin with the process rather than with the robot catalogue.

A practical design sequence is:

1. Define product variants, takt time, quality requirements, and production volume.
2. Map the process and identify handling, processing, inspection, and human tasks.
3. Select robot configuration, payload, reach, speed, and required accuracy.
4. Design fixtures and end effectors around part datum schemes and process forces.
5. Model reach, collisions, cable routing, ergonomics, maintenance access, and egress.
6. Specify sensors, machine interfaces, safety functions, and data collection.
7. Simulate cycle time and validate the proposed layout.
8. Build, integrate, test, and document the cell.
9. Conduct risk assessment, validation, operator training, and production ramp-up.

Cycle time is determined by the longest required production path, including robot motion, machine processing, loading, unloading, inspection, and waiting. If a cell produces one acceptable unit every 45 seconds, its theoretical hourly capacity is:

where C_h is units per hour and t_c is cycle time in seconds per unit. For 45 seconds, C_h = 3600/45 = 80 units/hour. Actual output is lower because of availability losses, performance losses, changeovers, rejects, and planned breaks.

Overall equipment effectiveness is commonly expressed as:

where A is availability, P is performance, and Q is quality, each expressed as a decimal. OEE is useful for diagnosing losses but should not be treated as the sole measure of business value.

6. Safety and Human–Machine Collaboration

Robot hazards include crushing, trapping, impact, unexpected restart, sharp tooling, hot surfaces, welding radiation, electrical energy, stored pneumatic energy, and hazardous substances. A safety strategy follows the hierarchy of controls: eliminate hazards where possible, use engineering safeguards, provide administrative controls and training, and use personal protective equipment as a final layer.

Traditional industrial robots normally operate inside a safeguarded space using fences, interlocked gates, light curtains, pressure-sensitive mats, scanners, emergency stops, and safety-rated monitoring. A safety function must be validated, not merely installed. Examples include safe torque off, safe limited speed, safe limited position, and monitored stop.

Collaborative operation is not synonymous with simply purchasing a collaborative robot. It is a mode of operation in which the robot and person can share a workspace under defined conditions. Common collaborative strategies include:

• Safety-rated monitored stop: the robot stops when a person enters.
• Hand guiding: an operator physically guides the robot using an enabling device.
• Speed and separation monitoring: robot speed or position changes according to human proximity.
• Power and force limiting: contact forces and energies are limited to acceptable values.

The application must be risk assessed because tooling, workpieces, sharp edges, clamping points, and robot motion can create hazards even when robot power is limited. Standards and legal requirements vary by jurisdiction; designers should consult applicable machinery-safety legislation and current standards such as ISO 10218 and ISO/TS 15066 where relevant. A risk assessment should identify hazards, estimate risk, implement controls, and verify residual risk. It should be repeated after changes in product, tooling, software, layout, or operating method.

Human–machine collaboration also involves cognitive and organisational issues. Poorly designed interfaces can cause alarm fatigue, overreliance on automation, loss of operator skill, or confusion about responsibility. Operators should understand system states, recovery procedures, safe zones, and escalation routes. Safety must remain effective during foreseeable misuse, maintenance, cleaning, and fault recovery—not only during normal production.

7. Applications

Handling and machine tending

Robots load and unload CNC machines, presses, injection moulders, and heat-treatment equipment. Benefits include reduced exposure to chips, heat, and repetitive lifting. Essential design elements are reliable part presentation, door and clamp interlocks, chip management, tool-life communication, and recovery from missed picks.

Assembly

Robotic assembly requires accurate part location, controlled insertion forces, orientation verification, and mistake-proofing. Compliance devices and force sensing can compensate for small positional errors. Assembly robots should confirm successful insertion rather than assuming that a commanded motion achieved the intended result.

Welding and joining

Robots support arc welding, spot welding, laser welding, and adhesive dispensing. Welding quality depends on torch angle, travel speed, stand-off distance, current, voltage, shielding, joint preparation, and thermal distortion. Seam tracking and vision can compensate for variation, but process qualification and inspection remain necessary.

Inspection

Vision-guided robots can position cameras, probes, or scanners. Inspection systems require controlled lighting, calibration artefacts, measurement uncertainty analysis, and clear acceptance criteria. A robot’s repeatability does not automatically guarantee measurement accuracy; the sensor, calibration, fixture, and algorithm all contribute to uncertainty.

8. Economics and Implementation Decisions

A robot investment should be assessed using total cost of ownership. Costs include robot and controller, tooling, fixtures, safety systems, integration, programming, training, facility modification, maintenance, energy, software, and downtime during installation. Benefits may include labour redeployment, greater output, reduced scrap, improved safety, lower ergonomic injury risk, and longer operating hours.

A simple payback estimate is:

This approximation assumes a constant annual benefit and ignores the time value of money. For larger investments, discounted cash-flow methods are preferable. Net present value is:

where I₀ is initial investment, CF_t is net cash flow in year t, r is the discount rate, and n is project life. Cash flows must be measured consistently, for example in currency per year, and assumptions about taxes, residual value, maintenance, inflation, and utilisation should be stated.

Automation is not automatically economical. Low volume, high product variety, unstable processes, frequent engineering changes, poor part presentation, or difficult maintenance can reduce value. Flexible fixtures, quick-change tooling, modular programming, and standard interfaces can improve the business case.

Applied Case: Illustrative Robotic Machine-Tending Cell

Background

A medium-sized manufacturer produces aluminium housings on two CNC machining centres. Operators manually load raw castings, remove finished components, and perform a visual check. The work is repetitive, involves chips and coolant, and causes occasional ergonomic complaints.

Problem

The company requires higher output and more consistent loading while retaining the ability to run three housing variants. The fixtures differ slightly, and raw castings vary in orientation.

Analysis

A six-axis articulated robot is selected because it can service both machines, rotate parts into fixture orientation, and place finished components at an inspection station. A two-jaw gripper with variant-specific inserts is considered, but vacuum gripping is rejected because casting surfaces are porous and contaminated with coolant. A vision sensor is placed above the infeed conveyor to identify orientation and variant. Fixture sensors confirm clamp status before machining begins.

The cell includes a guarded robot zone, interlocked access, a light curtain at the operator loading point, safe machine interfaces, chip-resistant cabling, and a manual recovery station. The operator loads a pallet and removes inspected parts without entering the robot’s active envelope.

Approach and solution

Offline simulation identifies a wrist singularity near one machine door, so the layout and approach path are changed. The robot uses joint motion between stations and linear motion for insertion into fixtures. The program requires positive confirmation of part presence, fixture clamp, machine door closure, and cycle completion. A reject chute prevents uncertain parts from entering the finished-goods flow.

The team calibrates the robot base, tool centre point, and vision frame. A pilot run measures cycle time, missed picks, fixture loading errors, and inspection rejects. Operators are trained in normal operation, fault recovery, lockout procedures, and escalation.

Outcome

The cell reduces manual exposure to coolant and repetitive lifting and provides traceability through part-variant and inspection records. Its success depends less on robot speed than on reliable part presentation, recovery logic, fixture sensing, and maintainability. The company retains an operator role for material replenishment, quality decisions, and process improvement.

Learning points

The case demonstrates that robot selection follows process requirements; sensing and fixtures are as important as the manipulator; collaborative operation is not necessary when physical separation is practical; and economic value depends on uptime, quality, and maintainability rather than purchase price alone.

9. Limitations, Governance, and Contemporary Developments

Robotic systems can amplify poor process design. Automation may transfer work rather than eliminate it, creating new demands for programming, troubleshooting, and maintenance skills. Cybersecurity is increasingly important because networked controllers, vision systems, and manufacturing execution systems can be attacked or misconfigured. Access control, network segmentation, software version control, backups, event logs, and secure remote support are essential.

Artificial intelligence is being applied to vision inspection, grasp planning, predictive maintenance, anomaly detection, and natural-language programming. These applications require representative data, validation, monitoring of false positives and false negatives, and human responsibility for consequential decisions. A machine-learning classifier should not be treated as infallible merely because its average accuracy is high; rare but costly failures require specific evaluation.

Digital twins and simulation support virtual commissioning, layout optimisation, and operator training. Mobile robots and autonomous mobile platforms increasingly transport materials between cells. Newer collaborative systems use improved force sensing, vision, adaptive planning, and safety-rated monitoring. Nevertheless, physical safeguards, validated safety functions, and competent engineering remain necessary.

Best-Practice Checklist

• Define the process, product variation, takt time, and quality criteria before selecting a robot.
• Separate robot accuracy, repeatability, payload, reach, and speed as different specifications.
• Design the tool, fixture, sensor, and robot as an integrated system.
• Use positive sensing and interlocks rather than relying on assumed sequence correctness.
• Simulate reach, collision, singularity, cycle time, ergonomics, and maintenance access.
• Conduct risk assessment for normal operation, foreseeable misuse, cleaning, and recovery.
• Validate safety functions and document changes.
• Measure availability, performance, quality, downtime causes, and maintenance burden.
• Train operators and maintenance staff in both routine use and safe recovery.
• Reassess cybersecurity, software versions, data governance, and human responsibility.

Chapter Summary

Industrial robots are programmable mechatronic systems whose configuration, kinematics, tooling, sensing, programming, and cell design must be selected together. Cartesian, SCARA, articulated, delta, and collaborative robots offer different combinations of reach, speed, stiffness, dexterity, and cost. Forward and inverse kinematics describe robot motion, while Jacobians support velocity, force, and singularity analysis. Accuracy and repeatability are distinct performance characteristics.

End effectors must retain parts under static and dynamic loads and must be integrated with sensing and fixtures. Effective programming includes motion, process logic, interlocks, error recovery, and data management. Cell design must address flow, cycle time, maintainability, safety, and human interaction. Collaborative applications require risk assessment and validated safety functions; a collaborative robot alone does not make an application safe. Economic evaluation should include total cost of ownership, utilisation, quality, maintenance, and lifecycle cash flows. Contemporary systems add vision, force control, digital twins, AI, mobile robotics, and connectivity, but these technologies require validation, cybersecurity, and accountable human oversight.

Key Terms

Degrees of freedom: Independent motions required to define a robot configuration.

End effector: Device attached to the robot wrist to grip, process, or inspect a workpiece.

Workspace: Set of positions and orientations reachable by a robot.

Singularity: Robot configuration in which motion capability is lost or joint velocities become problematic.

Forward kinematics: Calculation of tool pose from joint variables.

Inverse kinematics: Calculation of joint variables for a desired tool pose.

Jacobian: Matrix relating joint velocities to Cartesian velocities.

Accuracy: Closeness of actual position to the commanded or nominal position.

Repeatability: Closeness of repeated positions under the same conditions.

Payload: Maximum specified load the robot can handle under stated conditions.

Tool centre point: Reference point on the end effector used for programming.

Machine tending: Automated loading and unloading of manufacturing machines.

Teach pendant: Handheld interface used to program and operate a robot.

Offline programming: Robot programming and simulation performed away from the production cell.

Collaborative operation: Robot operation designed to permit defined human–robot interaction.

Risk assessment: Structured process for identifying hazards, estimating risk, and selecting controls.

OEE: Product of availability, performance, and quality factors.

Impedance control: Control that regulates the relationship between motion and contact force.

Digital twin: Digital representation used to model, monitor, or simulate a physical system.

Review Questions

5 Short Answer Questions

1. Distinguish robot accuracy from repeatability.
2. What is the function of a robot end effector?
3. Define a robot singularity.
4. Name two applications of machine vision in robotic manufacturing.
5. Why are interlocks important in a robotic cell?

5 Long Answer Questions

1. Compare Cartesian, SCARA, delta, and articulated robots in terms of structure and application.
2. Explain forward kinematics, inverse kinematics, and the role of homogeneous transformations.
3. Describe a systematic method for designing a robotic manufacturing cell.
4. Explain the principal collaborative operation strategies and their safety implications.
5. Discuss the technical and economic factors that determine whether automation is appropriate.

3 Analytical / Application Questions

1. A robot must lift a 6 kg part at 1.5 m/s² using two friction jaws with coefficient of friction 0.25 and safety factor 2.5. Calculate the minimum total normal force and discuss practical factors that may require a higher value.
2. Select a robot configuration for high-speed insertion of small components into a horizontal assembly fixture. Justify the selection and identify two limitations.
3. A proposed cell has excellent cycle time but frequent missed picks and difficult fault recovery. Analyse the likely system causes and recommend engineering and organisational improvements.

Multiple Choice Questions with Answers

Question 1

Which robot configuration is generally most suitable for high-speed planar assembly with a rigid vertical axis?

A. SCARA

B. Cartesian gantry

C. Spherical polar

D. Delta painting robot

Correct Answer: A. SCARA

Question 2

What does forward kinematics determine?

A. The required motor torque from friction

B. The tool pose from joint variables

C. The robot’s annual operating cost

D. The safety category of a cell

Correct Answer: B. The tool pose from joint variables

Question 3

A robot can repeatedly return to the same point but that point is offset from the commanded location. This describes:

A. High accuracy and low repeatability

B. Low accuracy and high repeatability

C. Low accuracy and low repeatability

D. A singularity necessarily occurring

Correct Answer: B. Low accuracy and high repeatability

Question 4

In the simplified friction-grip model, increasing the coefficient of friction while holding all other variables constant will:

A. Increase the required normal force

B. Decrease the required normal force

C. Eliminate the effect of acceleration

D. Reduce the part mass

Correct Answer: B. Decrease the required normal force

Question 5

Which function confirms that a robot program has achieved the intended physical operation rather than merely issued a motion command?

A. Positive sensing and process interlocking

B. Increasing the robot’s colour contrast

C. Removing all fixture sensors

D. Using joint motion for every path

Correct Answer: A. Positive sensing and process interlocking

Question 6

Which statement about collaborative robots is correct?

A. They are safe in every application without risk assessment

B. Their use can eliminate the need for guarding in all cases

C. The complete application, including tooling and workpiece, must be assessed

D. They cannot use vision or force sensing

Correct Answer: C. The complete application, including tooling and workpiece, must be assessed

Question 7

If a cell cycle time is 60 seconds per unit, its theoretical hourly capacity is:

A. 30 units/hour

B. 60 units/hour

C. 120 units/hour

D. 360 units/hour

Correct Answer: B. 60 units/hour

Question 8

Which factor is most directly associated with a robot singularity?

A. Loss of one or more Cartesian motion directions

B. An increase in product variety

C. A reduction in lighting intensity

D. A higher coefficient of friction

Correct Answer: A. Loss of one or more Cartesian motion directions

Question 9

Which item is normally included in total cost of ownership?

A. Only the robot purchase price

B. Only operator wages

C. Integration, tooling, maintenance, and energy costs

D. Only the resale value of the robot

Correct Answer: C. Integration, tooling, maintenance, and energy costs

Question 10

What is the primary purpose of a Jacobian in robot analysis?

A. To relate joint velocities to Cartesian velocities

B. To calculate annual depreciation

C. To classify welding defects visually

D. To replace all safety devices

Correct Answer: A. To relate joint velocities to Cartesian velocities