robot-sensors

name	robot-sensors
description	Robot sensors — encoders (incremental/absolute, resolution, accuracy), force-torque sensors (ATI, Schunk, 6-DOF wrench measurement), tactile sensors (capacitive, piezoresistive arrays), proximity sensors (laser rangefinder, ultrasonic, capacitive), vision systems (2D camera, 3D structured light, stereo, time-of-flight), IMU (MEMS accelerometer, gyroscope, Kalman filter), proprioception vs. exteroception, sensor fusion, ROS sensor interfaces, and safety sensors (ISO 13849).
metadata	{"priority":7,"promptSignals":{"phrases":["robot sensors","robot force sensor","robot encoder","force torque sensor","robot vision","tactile sensor robot"],"minScore":3}}

Robot Sensors — Complete Skill

Proprioceptive Sensors (Internal State)

Encoders (Position and Velocity)

Incremental encoder: Optical disc with slots → photodetector counts pulses; quadrature (A,B channels) → direction + count Resolution: N_ppr (pulses per revolution); with quadrature: 4 × N_ppr counts/revolution Velocity: ω = (N_counts / N_ppr) / Δt [rad/s; Δt = time window] Index channel (Z): one pulse per revolution → reference position Typical: 1,000–65,536 ppr for robot joints; 4,096 ppr (12-bit equivalent) for medium precision

Absolute encoder: Each position = unique binary code → no reference homing needed; no loss of position on power-off Types: single-turn (17–23 bit; 131,072–8,388,608 counts/rev); multi-turn (adds 12–16 bits of revolution count) Interfaces: SSI (synchronous serial), BiSS-C, EnDat 2.2, HIPERFACE Industrial robots: typically 17-23 bit single-turn + 16-bit multi-turn absolute

Resolution vs. accuracy: Resolution (encoder): Δθ_enc = 360°/counts_per_rev (perfect reading) Accuracy: limited by eccentricity, disc errors, interpolation errors; typically ±1–5 encoder counts

Motor-to-joint transmission: Joint position: θ_joint = θ_motor / i_gear [i_gear = gear ratio, e.g., 100:1] Joint resolution: Δθ_joint = Δθ_motor / i_gear → much finer than motor encoder alone But: backlash and compliance of gear reduces effective joint accuracy

Motor Current (Torque Sensing)

Torque estimation: τ_motor = k_T × i_motor [k_T = motor torque constant [Nm/A]; i_motor = current from drive] τ_joint = τ_motor × i_gear × η_gear - τ_friction [η_gear = gear efficiency; τ_friction = friction model] Accuracy: ±10–30% of rated torque due to friction uncertainty; useful for threshold detection

Application: torque monitoring for collision detection (abnormal torque → stop); approximate payload estimation

Force-Torque Sensors

6-Axis Wrench Measurement

ATI Gamma, Delta, Omega; Schunk FT-Axia; OnRobot HEX: Measures: 3 forces (Fx, Fy, Fz) and 3 torques (Mx, My, Mz) simultaneously Principle: bonded strain gauges on elastomeric structure; Wheatstone bridge Output: 6-channel analog or digital (Ethernet, EtherCAT, CAN)

Key specs: Gamma SI-130-10: Fxy ±130 N; Fz ±400 N; Txy ±10 Nm; Tz ±10 Nm; resolution 1/40 N, 1/800 Nm ATI Delta SI-330-30: ±330 N / ±30 Nm; heavier duty; assembly applications Stiffness: 10⁶–10⁷ N/m; natural frequency 800–1,500 Hz (unloaded)

Calibration matrix: F_meas = C × V_gauge [C = 6×6 calibration matrix; V_gauge = 6 strain voltages] Supplied by ATI with each sensor; re-calibration traceable to NIST force standards

Bias removal: τ_net = τ_measured - τ_gravity(q) - τ_bias [τ_gravity = gravity effect of tool weight; τ_bias = zero-load offset] Gravity compensation requires: mass of tool m_tool, CoM position d_c, robot configuration q

Applications: force-controlled assembly (peg-in-hole), surface grinding, polishing, human-robot interaction

Tactile Sensor Arrays

Capacitive arrays (BioTac, SynTouch; Tekscan FlexiForce): Matrix of capacitive elements; measures pressure distribution Resolution: 1–10 mm spatial; 0.1–10 N/cm² pressure Output: 2D force map → grasp quality assessment, slip detection, texture recognition

Piezoresistive arrays: Conductive polymer or ink changes resistance with pressure; cheaper; larger hysteresis Applications: robotic gripper fingertips; prosthetic hands

Exteroceptive Sensors (Environment)

Vision Systems

2D Camera: Industrial: 1–20 MP; GigE Vision or USB3; monochrome or color; frame rate 30–300 FPS Calibration: OpenCV (Zhang method); intrinsics (focal length f_x, f_y; principal point c_x, c_y; distortion k1,k2,p1,p2) Reprojection error: ≤ 0.5 pixels for good calibration

3D Structured Light (Keyence LJ-X, Cognex 3D-A; Photoneo PhoXi): Project pattern (fringe, binary code, random dots); 3D from phase shift or stereo triangulation Accuracy: 0.01–0.5 mm; speed: 0.1–10 Hz Applications: bin picking (depalletizing); weld seam tracking; part inspection

Stereo Vision: Two cameras + disparity map → depth z = f × B / d [f = focal length; B = baseline; d = disparity pixels] Accuracy: z_error = f × B × σ_d / d² [grows with z²; σ_d = disparity uncertainty ≈ 0.5–2 pixels] For z = 1 m, f = 1,000 px, B = 0.15 m, σ_d = 1 px: z_error = 0.15 mm

Time-of-Flight (Intel RealSense L515, Basler blaze): Laser or LED modulated light → measure phase shift of reflected signal → depth Accuracy: ±5–20 mm; frame rate: 30–90 Hz; range: 0.3–10 m Not suitable for specular surfaces (no reflection) or bright outdoor (sunlight interference)

LiDAR (Hokuyo UTM-30LX, Velodyne VLP-16): Rotating laser scanner; 2D or 3D; range 10–100 m; accuracy ±10–30 mm Used for mobile robot navigation; not for precision manipulation

Proximity Sensors

Ultrasonic (Pepperl+Fuchs UB; Banner Q50): Range: 0.03–10 m; accuracy ±1–5 mm; frequency 40–200 kHz; beam angle 10–25° Not affected by color/opacity; works in fog; poor for soft absorbing materials Applications: simple presence detection; level measurement; coarse distance

Laser rangefinder (Keyence LK, Micro-Epsilon optoNCDT): Range: 5–300 mm; accuracy ±0.01–0.1 mm; linearized output Principle: triangulation or time-of-flight (short range: triangulation preferred) Applications: part height measurement, thickness, surface profile

Inductive proximity (SICK, Balluff): Detects metallic objects; range 1–50 mm; no contact required; solid-state; O(μs) response Applications: end-of-stroke detection, part presence

Inertial Measurement Unit (IMU)

MEMS IMU (6-DOF or 9-DOF)

Accelerometer: Measures: linear acceleration (gravity + dynamics); MEMS spring-mass system Noise: 50–200 μg/√Hz (low-cost); 3–10 μg/√Hz (tactical grade) Bias stability: 0.1–10 mg (MEMS); 1–50 μg (navigation grade)

Gyroscope: Measures: angular velocity; MEMS Coriolis effect Noise: 0.01–0.1 °/s/√Hz; bias stability: 0.1–10 °/hr (MEMS); 0.001 °/hr (fiber optic)

Sensor fusion (complementary filter / Kalman): orientation = integral(gyro) + correction_from_accelerometer_when_nearly_static Kalman filter: state = [angle, gyro_bias]; measurement = accel direction (gravity vector); update attitude Madgwick / Mahony filters: computationally efficient quaternion-based orientation estimation

IMU applications in robotics: Mobile robot orientation; exoskeleton posture; UAV attitude; human motion capture (Xsens MVN) Not used for absolute joint position in fixed-base robots (encoders more accurate)

ROS Sensor Interfaces

ROS 2 standard message types: sensor_msgs/JointState: position, velocity, effort per joint sensor_msgs/Image: camera frames sensor_msgs/PointCloud2: 3D point cloud from depth camera or LiDAR geometry_msgs/WrenchStamped: force-torque sensor output sensor_msgs/Imu: IMU data (orientation, angular velocity, linear acceleration)

Calibration packages: camera_calibration (ROS): OpenCV Zhang method; monocular and stereo robot_calibration: joint-camera extrinsic calibration ft_calib: force-torque sensor gravity compensation calibration

Safety Sensors

Functional Safety (ISO 13849 / IEC 62061)

Safety LiDAR (SICK S3000, Pilz PSENscan): Monitors protective zone around robot; stops robot if human enters PLd/SIL2 rated; 2-channel safety output Applications: collaborative robots, AGVs, robot cell access monitoring

Safety light curtains: Detect hand/finger intrusion; PLe/SIL3 achievable Resolution: finger (14 mm) or body (30 mm) detection

Safety PLC (Pilz PNOZmulti, Siemens SIRIUS 3SK): Processes safety sensor signals; executes safety function (stop, reduce speed, safe torque off) Must be certified per ISO 13849 PL d or e

Standards and References

Standard	Scope
ISO 13849-1	Safety of machinery; control system safety (PL a–e)
IEC 62061	Safety of machinery; functional safety (SIL 1–3)
ISO 9283	Robot performance testing (indirectly references measurement sensors)
ANSI/A3 TR R15.606	Collaborative robot testing criteria
GigE Vision (AIA)	Camera interface standard
ROS REP 103/105	ROS coordinate system and unit conventions

Output

Provide: sensor selection table (sensor type; model; parameter measured; range; accuracy; interface; application within robot system), encoder spec (incremental/absolute; bits resolution; joint resolution after gear [°/count]; interface: EnDat/SSI), force-torque sensor (model; Fx/Fy/Fz range [N]; Mx/My/Mz range [Nm]; resolution; ROS driver), vision system (2D/3D; FOV; resolution; accuracy [mm]; lighting; calibration status), safety sensors (zone size [m]; PL/SIL rating; integration with robot controller E-stop chain), sensor fusion (if applicable: Kalman filter state; measurement update rate [Hz]; expected orientation accuracy [°]), ROS topics and message types for each sensor, calibration requirements (camera intrinsics; hand-eye calibration; FT zero procedure), and applicable standard (ISO 13849, GigE Vision, ROS REP 103).