Unit 6: Robotic Sensor Applications and Use cases from Industry

I. Orientation

Robotic sensors convert physical, chemical, or biological conditions into signals that a robot can measure and use for perception, control, safety, and decision-making. A complete robotic sensing system contains a sensing element, signal-conditioning circuit, data-processing method, and control response. Sensor performance is commonly evaluated through range, resolution, sensitivity, accuracy, repeatability, response time, robustness, and calibration stability.

• Measurement chain: The measured variable is converted into an electrical signal, filtered, digitized, interpreted, and supplied to the robot controller.
• Exteroception: Sensors such as cameras, lidar, radar, and microphones measure the external environment.
• Proprioception: Encoders, motor-current sensors, and joint torque sensors measure the robot’s internal state.
• Contact sensing: Tactile, force, and torque sensors detect physical interaction with objects or people.
• Closed-loop control: Sensor feedback allows the controller to compare desired and actual states and correct the robot’s motion.
• Uncertainty: Noise, drift, occlusion, delay, calibration error, and environmental changes affect every sensor measurement.
• Sensor fusion: Combining complementary sensors, such as lidar and inertial measurement units, improves reliability and reduces ambiguity.
• Safety principle: A sensor is useful only when its output leads to an appropriate action, such as stopping, slowing, repositioning, or rejecting an object.

II. Tactile and Soft Robotic Systems

These systems use deformable or contact-sensitive structures to interact safely with uncertain objects and environments. Their central requirement is to measure contact location, pressure, force, shape, or deformation without excessively restricting compliance.

A. Use cases in tactile sensing

Tactile sensing gives a robot information that cannot be obtained reliably from vision alone, especially during grasping and contact-rich manipulation.

• Contact detection: Resistive, capacitive, piezoresistive, piezoelectric, and optical taxels detect whether and where contact occurs. A taxel is an individual tactile sensing element in an array.
• Force distribution: A tactile array measures pressure across a gripper surface, helping the robot identify whether an object is centered or slipping.
• Slip prevention: High-frequency changes in shear force or vibration indicate object movement. The controller can increase grip force before the object falls.
• Object recognition: Tactile patterns reveal hardness, texture, curvature, and material properties when objects look visually similar.
• Assembly: Force-sensitive fingertips detect insertion contact in tasks such as placing a pin into a hole, preventing excessive lateral loads.
• Worked relationship: Pressure is calculated from normal force and contact area:

p = F / A

Here, p is pressure in pascals, F is normal force in newtons, and A is contact area in square metres. A 10 N force over 0.002 m² produces 5000 Pa.

• Limitations: Tactile sensors can suffer from hysteresis, abrasion, temperature dependence, wiring complexity, and calibration drift. Protective skins improve durability but may reduce spatial resolution.

B. Soft robotics

Soft robotics uses compliant materials such as silicone elastomers, fabrics, pneumatic chambers, and flexible polymers to produce adaptable motion and safer physical interaction.

• Deformation sensing: Stretchable resistive tracks, liquid-metal channels, fiber-optic sensors, and embedded strain gauges estimate bending or elongation.
• Pneumatic control: Pressure sensors measure air pressure inside soft actuators; a controller regulates valves to achieve a desired bend or gripping force.
• Shape estimation: Multiple distributed strain measurements can reconstruct the actuator’s curve rather than treating it as a rigid link.
• Human interaction: Soft grippers conform to fragile produce, food, biological tissue, and irregular household objects with lower damage risk.
• Medical devices: Soft wearable robots can assist movement while pressure and strain sensors monitor user comfort and applied force.
• Design trade-off: Compliance increases adaptability and safety, but soft bodies have nonlinear dynamics, material hysteresis, and less precise positioning than rigid mechanisms.
• Control requirement: Sensor feedback is essential because the same valve pressure may produce different shapes when payload, temperature, or actuator wear changes.

III. Industrial and Mobile Robot Sensing

Industrial robots require sensors for process quality and repeatable positioning, while mobile robots require sensors to estimate location and avoid collisions in changing environments.

A. Robotic arc welding sensors

Robotic arc welding sensors maintain weld quality by detecting joint position, torch distance, process stability, and defects during welding.

• Through-arc sensing: The controller observes welding current or voltage while the torch oscillates across a joint. Changes in arc length alter electrical measurements and indicate lateral tracking error.
• Laser seam tracking: A laser projects a line or point onto the workpiece. A camera detects the reflected profile and computes the seam center before or during welding.
• Arc-length control: Voltage is related to arc length in the operating range. A sudden voltage change can cause the robot to adjust torch height.
• Process monitoring: Current, voltage, wire-feed speed, gas flow, temperature, and acoustic signals reveal unstable transfer, lack of fusion, porosity, or burn-through.
• Torch geometry: A six-axis force/torque sensor can detect collision, contact, or abnormal resistance during approach and workpiece fitting.
• Worked example: If the measured seam center shifts 2 mm left during a programmed weave, the controller can add a compensating 2 mm rightward offset while maintaining travel speed.
• Industrial value: Sensors reduce dependence on perfectly fixtured parts, support adaptive welding, improve bead consistency, and reduce rework.
• Limitations: Welding produces intense light, heat, spatter, electromagnetic interference, and smoke, so sensors need shielding, filtering, and regular calibration.

B. Use cases of sensor in robot navigation

Navigation sensors allow a mobile robot to estimate its pose, build a map, plan a path, and avoid static or moving obstacles.

• Range measurement: Lidar measures distance using laser time-of-flight or phase shift; ultrasonic sensors use acoustic echoes and are useful at short range.
• Vision: RGB cameras provide appearance and object information, while depth cameras estimate three-dimensional structure.
• Inertial sensing: An IMU combines accelerometers and gyroscopes to measure linear acceleration and angular velocity. It supports short-term motion estimation.
• Wheel odometry: Encoders measure wheel rotation, but errors accumulate through wheel slip, uneven floors, and incorrect wheel-radius assumptions.
• Localization: Simultaneous localization and mapping, or SLAM, estimates robot pose while constructing or updating an environmental map.
• Sensor fusion: An extended Kalman filter or factor-graph method can combine lidar, camera, IMU, and odometry to reduce individual sensor weaknesses.
• Safety response: Proximity sensors trigger braking or speed reduction when an obstacle enters a protected zone.
• Environmental limitation: Dust, rain, reflective surfaces, darkness, transparent objects, and crowded spaces can reduce sensor reliability.

IV. Chemical, Wearable, and Agricultural Sensing

These applications extend robotics beyond mechanical perception by measuring substances, physiological states, and biological or environmental conditions.

A. Chemical sensing

Chemical sensors identify or quantify gases, vapors, liquids, or dissolved compounds through a selective physical or chemical reaction.

• Gas detection: Metal-oxide semiconductor sensors change resistance when gases such as carbon monoxide or methane interact with a heated surface.
• Electrochemical sensing: An analyte participates in an oxidation-reduction reaction at an electrode, producing a current related to concentration.
• pH measurement: A glass electrode develops a voltage that varies with hydrogen-ion activity; pH is defined as -log10 of hydrogen-ion activity.
• Electronic noses: Arrays of partially selective gas sensors generate a response pattern, which machine-learning models classify as a smell or contamination condition.
• Robotic use: Inspection robots detect leaks, hazardous fumes, explosive atmospheres, water pollutants, or food spoilage without exposing workers.
• Limitations: Selectivity, humidity dependence, sensor poisoning, response delay, and the need for calibration against known concentrations affect accuracy.

B. Use cases in wearables

Wearable sensors collect physiological and motion data from a person while maintaining comfort, low power consumption, and reliable skin contact.

• Motion monitoring: Accelerometers and gyroscopes in smart garments or exoskeletons estimate posture, gait, tremor, and limb movement.
• Physiological sensing: Photoplethysmography estimates pulse from changes in reflected light; skin-temperature, respiration, and galvanic skin-response sensors provide additional indicators.
• Human-robot cooperation: Wearable inertial units or force sensors communicate a worker’s movement and intention to a collaborative robot.
• Rehabilitation: Pressure insoles and joint-angle sensors measure weight distribution and range of motion during therapy.
• Exoskeleton control: Surface electromyography detects electrical activity from muscles, allowing assistance to begin when a user voluntarily initiates movement.
• Practical constraints: Motion artifacts, sweat, variable skin contact, battery life, wireless latency, and privacy protection must be addressed.

C. Use cases of sensors in agriculture

Agricultural robots use sensors to measure crop condition, soil properties, field geometry, and weather so that operations can be targeted rather than uniform.

• Crop imaging: RGB, multispectral, and hyperspectral cameras identify plant rows, weeds, disease symptoms, and nutrient stress.
• Vegetation indices: Normalized Difference Vegetation Index is commonly calculated as:

NDVI = (NIR - Red) / (NIR + Red)

NIR is near-infrared reflectance and Red is red-band reflectance. Higher values often indicate healthier vegetation, subject to crop and scene conditions.

• Soil sensing: Moisture, electrical conductivity, temperature, and nutrient sensors support irrigation and fertilization decisions.
• Precision spraying: Cameras detect individual weeds and actuate nozzles only where treatment is required, reducing chemical use.
• Harvesting: Force, color, depth, and tactile sensors help a robot locate ripe fruit and detach it without bruising.
• Field navigation: GNSS, lidar, cameras, and wheel encoders guide autonomous tractors or weeding robots between crop rows.
• Limitations: Mud, dust, changing sunlight, occlusion by leaves, uneven terrain, and seasonal variation complicate sensing.

V. Military and Healthcare Applications

In military and healthcare settings, sensing supports high-consequence decisions. Reliability, fail-safe behavior, data security, and human oversight are therefore as important as raw measurement accuracy.

A. Use cases in military

Military robots use sensors for reconnaissance, explosive-ordnance disposal, logistics, surveillance, and operation in hazardous environments.

• Reconnaissance: Daylight cameras, thermal imagers, lidar, and radar detect people, vehicles, terrain, and obstacles under varying visibility.
• Explosive detection: Chemical sensors identify explosive vapors, while ground-penetrating radar and metal detectors support mine and buried-object detection.
• Navigation: Inertial sensors, GNSS, lidar, and visual odometry maintain localization when satellite signals are weak or unavailable.
• Unmanned aerial systems: IMUs, barometers, magnetometers, optical-flow sensors, and cameras stabilize flight and support obstacle avoidance.
• Robotic manipulation: Force/torque sensors allow a disposal robot to handle suspicious objects while limiting contact force.
• Communications and security: Sensor data must be authenticated and protected because spoofed positioning or manipulated imagery can cause unsafe actions.
• Operational constraints: Vibration, shock, dust, weather, electromagnetic interference, limited power, and deliberate concealment challenge sensor performance.

B. Use cases in healthcare

Healthcare robots apply sensors to diagnosis, patient assistance, rehabilitation, surgery, and hospital logistics while preserving safety and clinical accountability.

• Surgical robotics: Endoscopic cameras provide visual feedback, while force sensors or instrument-tissue interaction estimates help prevent excessive tissue loading.
• Patient monitoring: Pulse oximeters estimate oxygen saturation from red and infrared light; ECG electrodes measure cardiac electrical activity.
• Rehabilitation: Encoders, pressure platforms, inertial sensors, and electromyography quantify movement and muscle activation.
• Assistive robots: Proximity, tactile, and force sensors detect patient contact and enable compliant operation during lifting or feeding.
• Mobile hospital robots: Lidar, cameras, ultrasonic sensors, and bump sensors support autonomous delivery of medicines, samples, and equipment.
• Sensor fusion: Combining camera-based pose estimation with wearable inertial data improves tracking when a body part is temporarily occluded.
• Safety and ethics: Medical sensors require calibration, redundancy, alarm validation, infection-control compatibility, informed consent, and protection of personal health data.
• Clinical limitation: A sensor reading is not automatically a diagnosis; validated algorithms and trained professionals must interpret measurements within clinical context.