**Robotics** is the discipline of machines that sense, decide and move on their own account, and this portal is its front door on Wikitube: article face [[Robotics]]. Thirty sections in five parts carry a reader from where the end of an arm is (kinematics), through what moves it and keeps it balanced (dynamics and actuation), what it believes about the world and itself (control and estimation), how it gets from here to there (mobile robots and navigation), to where it earns its keep and works alongside people (industry, interaction and learning). Every section has one Main article and one framework microsim; the microsims are further along than the articles right now, so most sections below carry a short stub instead of a finished page — see the Overview. Sibling spine: [[PORTAL_Engineering]], which this spine crosses at fourteen points. Index: [[PORTAL_INDEX]] · hub: [[PORTAL_Centers_of_Excellence]]. ## The hero: Humanoid robot <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/robotics/Humanoid_robot.html" data-title="Humanoid robot"></div> *Try: run the tour stops in order (kinematics → balance → sensing → navigation → anatomy); then jump straight to a part with `?part=Rotary_encoder&embed=1` in the URL, or any of the thirty-odd focusable parts named in the Work list below.* About thirty of the hero's parts are individually focusable, and each names the Engineering or Robotics section that explains it — the neck lidar ring opens [[#Sensors|Sensors]], the hip actuator opens [[#Actuators|Actuators]], the leg's inverse-kinematics solve opens [[#Inverse kinematics|Inverse kinematics]], and so on through fifteen mechanical parts shared with [[PORTAL_Engineering]] and fourteen sensing, actuation and estimation parts that are this portal's own. It is not modelled on any existing character, product or brand. It walks, squats, waves and idles on a lit stage; its inverted-pendulum pelvis keeps the zero moment point inside its soles with better than a 2.3 cm margin at every speed it demonstrates; an exploded anatomy view pulls it apart with leader lines naming every part; a five-stop guided tour covers kinematics, balance, sensing, navigation and anatomy in order. **On the spine:** [[Robotics]] · [[Robotics_engineering]] · [[Mechatronics]] · sibling spine [[PORTAL_Engineering]] · hub [[PORTAL_Centers_of_Excellence]]. ## Part I — Kinematics: where the end of the arm is *Serves the [[Robotics]], [[Robotics_engineering]] and [[Mechanical_engineering]] shells.* Before a robot can be controlled it has to be described: where its end effector sits given its joint angles (forward kinematics), which joint angles reach a given point (inverse kinematics), how many independent motions its mechanism actually has, and what shape of workspace a serial or parallel arm sweeps out. ### Forward kinematics *Main article: — stub, article not yet written (W4) · See also: [[Forward_kinematics]], [[Denavit–Hartenberg_parameters]], [[Kinematic_chain]]* > **Stub.** *Robot kinematics* has no Wikitube article yet. A three-link planar arm's end-effector position and > orientation follow from chaining a transform per joint — the Denavit–Hartenberg convention names four parameters > per joint (link length, twist, offset, angle) and multiplies four matrices to get there, the same table printed > beside the arm and updating live as the joint sliders move. <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/robotics/Robot_kinematics.html" data-title="Robot kinematics"></div> *Try: move the three joint sliders and watch the end-effector position and the DH table update together.* Connects to: [[#Inverse kinematics|Inverse kinematics]] · [[#Degrees of freedom and mobility|Degrees of freedom]] · [[#Serial manipulators and workspace|Serial manipulators]] · [[PORTAL_Engineering#The four-bar linkage|Four-bar linkage]] ### Inverse kinematics *Main article: — stub, article not yet written (W4) · See also: [[Jacobian_matrix_and_determinant]], [[Robotic_arm]]* > **Stub.** *Inverse kinematics* has no Wikitube article yet. Drag a target and a 2-link arm's analytic solution > shows both its branches at once — elbow-up and elbow-down — plus the reach circle it cannot cross and the full > stretch it hits at the workspace boundary; a 3-link arm switches to the iterative route instead, a Jacobian > pseudo-inverse nudging the joints toward the target step by step, which is what most real arms actually run. <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/robotics/Inverse_kinematics.html" data-title="Inverse kinematics"></div> *Try: drag the target near the edge of reach and watch the elbow-up and elbow-down branches converge; switch to the 3-link Jacobian iteration and watch it walk toward the target one step at a time.* Connects to: [[#Forward kinematics|Forward kinematics]] · [[#Serial manipulators and workspace|Serial manipulators]] · [[#Parallel manipulators|Parallel manipulators]] · [[#Manipulator dynamics|Manipulator dynamics]] ### Degrees of freedom and mobility *Main article: — stub, article not yet written (W4) · See also: [[Chebychev–Grübler–Kutzbach_criterion]], [[Kinematic_pair]]* > **Stub.** *Degrees of freedom (mechanics)* has no Wikitube article yet. Build a planar chain one link at a time — > revolute, prismatic, or a closed loop — and the Grübler–Kutzbach formula M = 3(L − 1) − 2J − H counts how many > independent motions it has left: a triangle of three links locks solid at M = 0, a four-bar needs exactly one > motor, a five-bar needs two, and the sim builds each case link by link so the count is never just asserted. <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/robotics/Degrees_of_freedom_(mechanics).html" data-title="Degrees of freedom (mechanics)"></div> *Try: add links one at a time and watch M count down; close a loop into a four-bar and read M land on exactly one.* Connects to: [[#Forward kinematics|Forward kinematics]] · [[#Serial manipulators and workspace|Serial manipulators]] · [[#Parallel manipulators|Parallel manipulators]] · [[PORTAL_Engineering#The four-bar linkage|Four-bar linkage]] ### Serial manipulators and their workspace *Main article: — stub, article not yet written (W4) · See also: [[Articulated_robot]], [[SCARA]], [[Manipulator_(device)]]* > **Stub.** *Serial manipulator* has no Wikitube article yet. Sweep a 2R or 3R arm through its joint limits and the > shaded region it can reach is its workspace; inside that region a smaller, dexterous subset is where the arm can > also point its end effector any direction it likes, not just reach the point. A SCARA arm (two revolute joints in > a plane plus a vertical stroke) trades that full dexterity for speed and stiffness in one direction, against a > fully articulated arm's more general but slower reach. <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/robotics/Serial_manipulator.html" data-title="Serial manipulator"></div> *Try: sweep the joint limits and watch the shaded workspace fill in; switch from articulated to SCARA and compare the shapes.* Connects to: [[#Forward kinematics|Forward kinematics]] · [[#Inverse kinematics|Inverse kinematics]] · [[#Parallel manipulators|Parallel manipulators]] · [[#Industrial robots|Industrial robots]] ### Parallel manipulators *Main article: — stub, article not yet written (W4) · See also: [[Stewart_platform]], [[Delta_robot]], [[Hexapod_(robotics)]]* > **Stub.** *Parallel manipulator* has no Wikitube article yet — and its sim is a true 3-D scene. A Stewart > platform's six pose sliders (x, y, z, roll, pitch, yaw) drive an *inverse* kinematics solve for all six leg > lengths at once, each leg coloured by how far it has extended; push the pose far enough and a leg hits its stroke > limit and the workspace boundary shows up directly. A delta robot is the same closed-loop idea specialised for > speed rather than orientation, and a hexapod's walking gait ([[#Legged locomotion and gaits|Part II]]) reuses this > same platform as the body six legs carry. <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/robotics/Parallel_manipulator.html" data-title="Parallel manipulator"></div> *Try: orbit the 3-D scene while sliding roll and pitch and watch the six legs colour by extension; push a pose until one leg hits its stroke limit.* Connects to: [[#Inverse kinematics|Inverse kinematics]] · [[#Degrees of freedom and mobility|Degrees of freedom]] · [[#Serial manipulators and workspace|Serial manipulators]] · [[#Walking and the zero moment point|Zero moment point]] · [[#Legged locomotion and gaits|Legged locomotion]] ## Part II — Dynamics, actuation and balance *Serves the [[Electromechanics]], [[Control_engineering]] and [[Mechatronics]] shells, and Parts II–III of [[PORTAL_Engineering]].* Once a robot is described kinematically it has to be driven and kept upright: the motors, cylinders and screws that actually apply force, the servo loop wrapped around them, the equations of motion a loaded arm obeys, the inverted-pendulum problem that is balance in its simplest form, and the gait that keeps a legged robot's zero moment point inside its feet. ### Actuators *Main article: — stub, article not yet written (W4) · See also: [[Linear_actuator]], [[Pneumatic_actuator]]. Link only: [[Hydraulic_cylinder]] (Engineering ENG24)* > **Stub.** *Actuator* has no Wikitube article yet. One load, three actuator choices on a select: a DC motor's > speed falls in a straight line against torque, geared down to trade speed for force at the joint; a hydraulic > cylinder simply multiplies pressure by area, F = pA; a pneumatic cylinder behaves like a soft spring because the > gas itself compresses. Overlay all three against the same load line and the sim shows which one actually delivers > the needed force at the needed speed. <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/robotics/Actuator.html" data-title="Actuator"></div> *Try: step through DC-motor, hydraulic and pneumatic and watch each one's force–speed curve cross the same load line differently.* Connects to: [[#Servomechanisms|Servomechanisms]] · [[#Manipulator dynamics|Manipulator dynamics]] · [[#Aerial robots|Aerial robots]] · [[PORTAL_Engineering#Simple machines and mechanical advantage|Simple machines]] · [[PORTAL_Engineering#Gears and gear trains|Gears]] · [[PORTAL_Engineering#Hydraulics and pneumatics|Hydraulics]] ### Servomechanisms *Main article: — stub, article not yet written (W4) · See also: [[Servomotor]], [[Stepper_motor]], [[Rotary_encoder]], [[Synchro]]* > **Stub.** *Servomechanism* has no Wikitube article yet. A position servo — a motor inside a P or PD feedback > loop — steps to a new setpoint and its response shows the classic trade: raise the gain and it settles faster but > overshoots more, and the following error under a steady ramp never quite closes to zero on a P-only loop. Its > feedback comes from an encoder, and at low resolution that feedback itself becomes visibly staircased rather than > smooth, a quantisation the sim doesn't hide. <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/robotics/Servomechanism.html" data-title="Servomechanism"></div> *Try: raise the proportional gain and watch overshoot grow as settling time falls; then drop the encoder resolution and watch the feedback staircase.* Connects to: [[#Actuators|Actuators]] · [[#PID control|PID control]] · [[#Sensors|Sensors]] · [[#Industrial robots|Industrial robots]] · [[PORTAL_Engineering#Vibration isolation|Vibration isolation]] ### Manipulator dynamics *Main article: — stub, article not yet written (W4) · See also: [[Newton–Euler_equations]]. Link only: [[Moment_of_inertia]] (Engineering ENG13)* > **Stub.** *Lagrangian mechanics* has no Wikitube article yet. Swing a 2-link arm and its equations of motion split > cleanly into a gravity term, an inertia term and a Coriolis/centripetal term that couples the two joints; toggle > gravity compensation off and the arm sags exactly where the gravity term predicts, and raising the payload changes > every term at once, not just the one that "obviously" should change. The Newton–Euler form gets the same torques > a different way — a recursive per-link pass forward for velocities and accelerations, then backward for forces — > and the variant draws each link's force explicitly rather than folding them into one Lagrangian expression. <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/robotics/Lagrangian_mechanics.html" data-title="Lagrangian mechanics"></div> *Try: toggle gravity compensation off and watch the arm sag; raise the payload and read all three torque terms change together, not just the gravity one.* Connects to: [[#Inverse kinematics|Inverse kinematics]] · [[#Actuators|Actuators]] · [[#The inverted pendulum|Inverted pendulum]] · [[PORTAL_Engineering#Flywheels and inertia|Flywheels]] ### The inverted pendulum *Main article: — stub, article not yet written (W4) · See also: [[Self-balancing_scooter]], [[Linear–quadratic_regulator]], [[State-space_representation]]* > **Stub.** *Inverted pendulum* has no Wikitube article yet. A cart-pole balanced by linear state feedback, > u = −Kx, is the cleanest possible balance problem: four gains on four sliders, and getting even one sign wrong > drops the pole in seconds while the right combination rejects a disturbance kick and settles back upright. The > LQR variant derives those same four gains automatically from how much the controller is told to penalise error > versus effort (the Q and R weights), rather than hand-tuning them, and prints the state-space A and B matrices the > whole problem is built from. <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/robotics/Inverted_pendulum.html" data-title="Inverted pendulum"></div> *Try: flip one gain's sign and watch the pole fall; then switch to LQR and watch the same four gains emerge from just the Q and R weights.* Connects to: [[#Manipulator dynamics|Manipulator dynamics]] · [[#Walking and the zero moment point|Zero moment point]] · [[#PID control|PID control]] · [[#Sensors|Sensors]] ### Walking and the zero moment point *Main article: — stub, article not yet written (W4) · See also: [[Humanoid_robot]] (the portal's hero, above), [[Bipedalism]]* > **Stub.** *Zero moment point* has no Wikitube article yet. A sagittal biped's centre of mass rides a slider-driven > trajectory, and its acceleration alone determines where the zero moment point falls; balance holds exactly as > long as that point stays inside the support polygon — one foot's footprint in single support, the overlap of both > in double support — and is lost the instant it doesn't. The [[#The hero Humanoid robot|Humanoid_robot hero]] runs > this exact physics at full scale, its ZMP margin against the support polygon read out live as it walks. <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/robotics/Zero_moment_point.html" data-title="Zero moment point"></div> *Try: drive the CoM trajectory faster and watch the ZMP approach the edge of the support polygon; switch from double to single support and watch the margin shrink.* Connects to: [[#Parallel manipulators|Parallel manipulators]] · [[#The inverted pendulum|Inverted pendulum]] · [[#Wearable robots|Wearable robots]] · [[#Legged locomotion and gaits|Legged locomotion]] ### Legged locomotion and gaits *Main article: — stub, article not yet written (W4) · See also: [[Hexapod_(robotics)]] (walking default, shared with Part I), [[Central_pattern_generator]], [[Passive_dynamics]]* > **Stub.** *Legged robot* has no Wikitube article yet. A hexapod or quadruped's gait diagram — which feet are down > at which instant — falls out of just two numbers, a duty factor and a set of phase offsets between legs; switch > the select from tripod to trot to walk and the same generator produces every standard gait, while the support > polygon and a stability margin update with the body's own motion. A central pattern generator gets the same > rhythm from coupled phase oscillators rather than a lookup table, closer to how a real animal's spinal circuitry > actually produces it; a compass-gait walker shows the opposite extreme, a gait with no motors and no controller > at all, driven purely by gravity down a shallow slope. <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/robotics/Legged_robot.html" data-title="Legged robot"></div> *Try: step from tripod to trot to walk and watch the support polygon and stability margin change with it; switch to the passive compass gait and watch it walk with no motor at all.* Connects to: [[#Parallel manipulators|Parallel manipulators]] · [[#Walking and the zero moment point|Zero moment point]] · [[#Wearable robots|Wearable robots]] · [[#Braitenberg vehicles|Braitenberg vehicles]] · [[PORTAL_Engineering#Design for additive manufacturing|Additive manufacturing]] · [[PORTAL_Engineering#Material selection|Material selection]] ## Part III — Control and estimation *Serves [[Control_engineering]], [[Instrumentation_and_control_engineering]], [[PORTAL_Control_theory]], [[PORTAL_Feedback]] and [[PORTAL_Signal_Processing]].* Part III is what a robot runs on and believes: the PID loop most machines actually use, the Kalman filter that turns noisy measurements into a trustworthy estimate, reactive behaviour that needs no model of the world at all, the model-predictive controller that plans a horizon ahead, and the sensors every one of the other four depends on. ### PID control <!-- SECTIONSIMS:BEGIN g34 2026-09-19 - microsims the articles linked here play; generated by _tools/generate/g34_portal_section_sims.py; do not hand-edit inside --> *Microsims from the articles this section links:* <div class="wt-simrow"> <div class="wt-sim" data-src="https://editor.p5js.org/sciencenibber/full/VOi4_2NFp" data-title="PID controller"></div> <div class="wt-sim" data-src="https://editor.p5js.org/sciencenibber/full/DtAoGXBbj" data-title="Control theory"></div> <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/Feedback.html" data-title="Feedback"></div> <div class="wt-sim" data-src="https://editor.p5js.org/sciencenibber/full/sgQFSGMnw" data-title="Feedback · p5.js"></div> </div> <!-- SECTIONSIMS:END --> *Main article: [[PID_controller]] (thin legacy — promoted, not rewritten) · See also: [[Proportional_control]], [[Ziegler–Nichols_method]], [[Setpoint_(control_system)]], [[Closed-loop_controller]]. Link (exist): [[Control_theory]], [[Control_system]], [[Feedback]]* Three gains — proportional, integral, derivative — steer a first-order-plus-dead-time plant to a setpoint: the proportional term alone always leaves an offset it can never fully close, integral action erases that offset at the cost of overshoot, and derivative action damps the overshoot back down. A Ziegler–Nichols tune skips hand iteration entirely — push the gain up until the loop rings at a sustained, undamped cycle, read off that ultimate gain and period, and the classic rules fill in Kp, Ki and Kd from those two numbers alone. <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/robotics/PID_controller.html" data-title="PID controller"></div> *Try: run P-only and watch the steady offset that never closes; add integral action and watch it close at the cost of overshoot; press the Ziegler–Nichols button and watch all three gains fill in from one sustained-cycle test.* Connects to: [[#Servomechanisms|Servomechanisms]] · [[#The inverted pendulum|Inverted pendulum]] · [[#State estimation and the Kalman filter|Kalman filter]] · [[#Model predictive control|Model predictive control]] ### State estimation and the Kalman filter <!-- SECTIONSIMS:BEGIN g34 2026-09-19 - microsims the articles linked here play; generated by _tools/generate/g34_portal_section_sims.py; do not hand-edit inside --> *Microsims from the articles this section links:* <div class="wt-sim" data-src="https://editor.p5js.org/sciencenibber/full/CHkjVQYrl" data-title="Kalman filter"></div> <!-- SECTIONSIMS:END --> *Main article: [[Kalman_filter]] (thin legacy — promoted, not rewritten) · See also: [[Extended_Kalman_filter]], [[Sensor_fusion]]* A target moving at constant velocity is tracked through noisy range measurements alone, frame by frame: predict where it should be, compare against where the noisy measurement says it is (the innovation), weigh the two by a gain that the filter itself computes, and update — while its own uncertainty, drawn as a shrinking covariance ellipse, tightens as evidence accumulates. Raise the process noise relative to the measurement noise and the filter trusts the new reading more; raise it the other way and it trusts its own prediction more. Extend the same idea to a nonlinear measurement (range and bearing together) and it becomes the extended Kalman filter that underwrites [[#Localization and mapping|SLAM]]; combine an accelerometer and a gyroscope through it — or through the simpler complementary filter — and it becomes sensor fusion. <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/robotics/Kalman_filter.html" data-title="Kalman filter"></div> *Try: raise measurement noise and watch the covariance ellipse stay wide and the filter lean on its own prediction; lower it and watch the ellipse collapse onto the truth.* Connects to: [[#PID control|PID control]] · [[#Sensors|Sensors]] · [[#Mobile robots|Mobile robots]] · [[#Localization and mapping|Localization and mapping]] · [[#Monte Carlo localization|Monte Carlo localization]] ### Behavior-based control *Main article: — stub, article not yet written (W4) · See also: [[Behavior-based_robotics]]* > **Stub.** *Subsumption architecture* has no Wikitube article yet. A 2-D robot among obstacles runs three layered > behaviours at once — avoid, wander, seek light — each one able to subsume (override) the layer below it when it > has something urgent to say, with no central model of the world coordinating them. Switch a layer off and the > emergent behaviour changes qualitatively, not just in degree: without "avoid" the robot heads straight into > obstacles; without "seek light" it just wanders. <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/robotics/Subsumption_architecture.html" data-title="Subsumption architecture"></div> *Try: toggle each layer off in turn and watch which piece of the emergent behaviour disappears with it.* Connects to: [[#Mobile robots|Mobile robots]] · [[#Motion planning|Motion planning]] · [[#Swarms|Swarms]] · [[#Braitenberg vehicles|Braitenberg vehicles]] ### Braitenberg vehicles *Main article: — stub, article not yet written (W4) · See also: none minted; `Behavior-based_robotics` links here* > **Stub.** *Braitenberg vehicle* has no Wikitube article yet. Two light sensors wire straight to two wheels — no > processor, no model, just wires — and the wiring alone decides the vehicle's apparent temperament: crossed and > excitatory looks aggressive (2b), uncrossed and excitatory looks fearful (2a), crossed and inhibitory looks > loving (3a), uncrossed and inhibitory looks like curious exploration (3b). Drag the lamp and watch a machine with > no intelligence at all look, unmistakably, like it has some. <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/robotics/Braitenberg_vehicle.html" data-title="Braitenberg vehicle"></div> *Try: step through all four wirings with the same lamp position and watch the "personality" change with nothing but the wiring.* Connects to: [[#Behavior-based control|Behavior-based control]] · [[#Mobile robots|Mobile robots]] · [[#Swarms|Swarms]] · [[#Legged locomotion and gaits|Legged locomotion]] ### Model predictive control *Main article: [[Model_predictive_control]] (thin legacy — promoted, not rewritten) · See also: [[Optimal_control]], [[Trajectory_optimization]]* A double integrator — position and velocity, driven toward a target under a hard limit on the input — is planned one receding horizon at a time: at every step the controller solves for the best short plan it can under the limit, executes only the first move, then re-solves from wherever it actually ended up. The predicted trajectory redraws every step, which is the whole difference from a fixed open-loop plan (the Trajectory_optimization variant shows that plan alone, computed once and never revised) — MPC's advantage is exactly that it keeps replanning against reality instead of trusting its first guess. <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/robotics/Model_predictive_control.html" data-title="Model predictive control"></div> *Try: lengthen the horizon and watch the predicted trajectory smooth out; tighten the input limit and watch the controller take longer to reach the target rather than violate it.* Connects to: [[#The inverted pendulum|Inverted pendulum]] · [[#PID control|PID control]] · [[#Motion planning|Motion planning]] · [[#Self-driving cars|Self-driving cars]] ### Sensors <!-- SECTIONSIMS:BEGIN g34 2026-09-19 - microsims the articles linked here play; generated by _tools/generate/g34_portal_section_sims.py; do not hand-edit inside --> *Microsims from the articles this section links:* <div class="wt-sim" data-src="https://editor.p5js.org/sciencenibber/full/hmYLjqAjQ" data-title="Sensor"></div> <!-- SECTIONSIMS:END --> *Main article: [[Sensor]] (thin legacy — promoted, not rewritten) · See also: [[Accelerometer]], [[Gyroscope]], [[Inertial_measurement_unit]], [[Lidar]]. Variant of [[#Servomechanisms|Servomechanisms]]: [[Rotary_encoder]]* One true signal passes through a sensor model with five knobs that every real sensor has some of: bandwidth (a first-order lag that smooths fast changes away), resolution (quantisation into discrete steps), noise, bias and drift — each isolated on its own slider so the difference between accuracy (how close to the truth, on average) and precision (how tightly repeatable) shows up directly on a target plot instead of staying an abstract distinction. The IMU variant pairs a gyroscope's fast-but-drifting estimate with an accelerometer's slow-but-stable one, the exact trade [[#State estimation and the Kalman filter|sensor fusion]] exists to resolve; the lidar variant turns time-of-flight ranging and angular resolution into the same sensor-model language. <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/robotics/Sensor.html" data-title="Sensor"></div> *Try: add bias alone and watch the target plot go precise-but-wrong; add noise alone and watch it go accurate-but-scattered; then combine gyro drift and accelerometer noise in the IMU variant.* Connects to: [[#Servomechanisms|Servomechanisms]] · [[#State estimation and the Kalman filter|Kalman filter]] · [[#Localization and mapping|Localization and mapping]] · [[#Aerial robots|Aerial robots]] ## Part IV — Mobile robots and navigation *Serves [[Automotive_engineering]] (Engineering Part III), [[PORTAL_Graph_theory]], [[PORTAL_Agent-based_model]] and [[PORTAL_Self-organization]].* Part IV moves the robot: how a wheeled or omnidirectional base actually drives, how it plans a path through obstacles, how it figures out where it is while building a map of where it's been, and the two robots — the swarm and the single vehicle at speed — that stretch the same navigation problem to its extremes. ### Mobile robots *Main article: — stub, article not yet written (W4) · See also: [[Differential_wheeled_robot]], [[Robot_navigation]], [[Autonomous_robot]]* > **Stub.** *Mobile robot* has no Wikitube article yet. A differential-drive base's two wheel speeds set both its > forward velocity and its turn rate, and from those two numbers alone falls the instantaneous centre of rotation > the robot is actually pivoting around. Odometry integrates those same wheel counts over time to estimate position > — and drifts from the truth doing it, a gap the sim draws directly by running the true path and the odometry > estimate side by side until they visibly part ways. <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/robotics/Mobile_robot.html" data-title="Mobile robot"></div> *Try: set the two wheel speeds unequal and watch the turn radius and instantaneous centre of rotation appear; then watch the odometry trace drift from the true path over time.* Connects to: [[#State estimation and the Kalman filter|Kalman filter]] · [[#Omnidirectional wheels|Omnidirectional wheels]] · [[#Motion planning|Motion planning]] · [[#Localization and mapping|Localization and mapping]] · [[#Self-driving cars|Self-driving cars]] · [[PORTAL_Engineering#Steering geometry and handling|Steering geometry]] ### Omnidirectional wheels *Main article: — stub, article not yet written (W4) · See also: [[Omni_wheel]]* > **Stub.** *Mecanum wheel* has no Wikitube article yet. Four mecanum wheels, their rollers set at 45°, convert > independent wheel speeds into any combination of forward, sideways and rotating motion at once — drag a joystick > target and the sim solves backward for the four wheel-speed arrows that produce exactly that body twist (vx, vy, > ω), then forward again to confirm them. A three-wheel kiwi drive gets the same holonomic freedom from ordinary > rollers set at 120° instead of a mecanum roller pattern. <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/robotics/Mecanum_wheel.html" data-title="Mecanum wheel"></div> *Try: drag the joystick target sideways — motion an ordinary wheeled robot cannot make at all — and watch the four wheel-speed arrows that produce it.* Connects to: [[#Mobile robots|Mobile robots]] · [[#Industrial robots|Industrial robots]] · [[PORTAL_Engineering#Steering geometry and handling|Steering geometry]] ### Motion planning *Main article: — stub, article not yet written (W4) · See also: [[Rapidly_exploring_random_tree]], [[Probabilistic_roadmap]], [[A*_search_algorithm]], [[Any-angle_path_planning]], [[Obstacle_avoidance]]. See also (exists thin): [[Configuration_space_(physics)]]* > **Stub.** *Motion planning* has no Wikitube article yet. On a grid world with obstacles, A* expands its search > frontier one cell at a time toward the goal, guided by a heuristic that trades optimality for speed as its weight > rises; an RRT grows the same search by scattering random samples and connecting each to its nearest tree node > instead of exploring a grid, and the sim compares both algorithms' path length and node count side by side on the > identical obstacle field. A probabilistic roadmap samples the free space once and reuses it for many queries; > obstacle avoidance by potential field pushes the robot away from obstacles continuously rather than planning a > discrete path at all; and the configuration-space variant reframes a 2-link arm's own joint limits as obstacles in > its own C-space — the same planning problem, one dimension removed from the physical world. <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/robotics/Motion_planning.html" data-title="Motion planning"></div> *Try: watch A* expand its frontier toward the goal on the grid; switch to RRT on the same obstacle field and compare path length and node count.* Connects to: [[#Behavior-based control|Behavior-based control]] · [[#Model predictive control|Model predictive control]] · [[#Mobile robots|Mobile robots]] · [[#Localization and mapping|Localization and mapping]] · [[PORTAL_Engineering#Machining and CNC|Machining and CNC]] ### Localization and mapping *Main article: — stub, article not yet written (W4) · See also: [[Robotic_mapping]], [[Occupancy_grid_mapping]], [[Visual_odometry]]* > **Stub.** *Simultaneous localization and mapping* has no Wikitube article yet. A robot drives past a handful of > landmarks, ranging each one with noise, and EKF-SLAM tracks the robot's own pose ellipse and every landmark's > ellipse together in one filter; all of them grow with uncertainty as the robot drives, then — the payoff — snap > tighter together the instant the robot revisits a landmark it already knows, a loop closure that corrects years > of accumulated drift in one update. Occupancy-grid mapping builds the same picture from range scans instead of > discrete landmarks, accumulating log-odds of occupancy cell by cell. <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/robotics/Simultaneous_localization_and_mapping.html" data-title="Simultaneous localization and mapping"></div> *Try: drive past all the landmarks once and watch every ellipse grow; then close the loop back through a known landmark and watch them all snap tighter together at once.* Connects to: [[#State estimation and the Kalman filter|Kalman filter]] · [[#Sensors|Sensors]] · [[#Mobile robots|Mobile robots]] · [[#Motion planning|Motion planning]] · [[#Monte Carlo localization|Monte Carlo localization]] ### Monte Carlo localization *Main article: — stub, article not yet written (W4) · See also: [[Particle_filter]], [[Odometry]], [[Dead_reckoning]]* > **Stub.** *Monte Carlo localization* has no Wikitube article yet. On a known map with an unknown starting pose, a > cloud of hundreds of particles — each a guess at the robot's pose — is moved by the same odometry every step and > reweighted by how well a simulated range scan from each particle's position matches the real one; resampling > keeps the particles that fit and discards the ones that don't, and the whole cloud collapses onto the true pose > as evidence accumulates. A kidnapping button teleports the real robot without telling the filter, scattering the > cloud back out — the same recovery Monte Carlo localization is prized for that a pure Kalman filter cannot do. <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/robotics/Monte_Carlo_localization.html" data-title="Monte Carlo localization"></div> *Try: watch the particle cloud collapse onto the true pose as the robot drives; press kidnap and watch it scatter, then recover.* Connects to: [[#State estimation and the Kalman filter|Kalman filter]] · [[#Sensors|Sensors]] · [[#Mobile robots|Mobile robots]] · [[#Localization and mapping|Localization and mapping]] ### Swarms <!-- SECTIONSIMS:BEGIN g34 2026-09-19 - microsims the articles linked here play; generated by _tools/generate/g34_portal_section_sims.py; do not hand-edit inside --> *Microsims from the articles this section links:* <div class="wt-simrow"> <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/Emergence.html" data-title="Emergence"></div> <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/Self-organization.html" data-title="Self-organization"></div> </div> <!-- SECTIONSIMS:END --> *Main article: — stub, article not yet written (W4) · See also: [[Swarm_intelligence]], [[Boids]]. Link (exist, live sims): [[Emergence]], [[Self-organization]]* > **Stub.** *Swarm robotics* has no Wikitube article yet. Up to four hundred boids follow only three local rules — > separation, alignment, cohesion — each on its own weight slider, with no leader and no global plan, and the order > parameter read out live says how far the flock has organised itself from random motion toward one coherent > heading. Click to drop a predator into the flock and the same three rules produce evasive swirling with no fourth > rule added. <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/robotics/Swarm_robotics.html" data-title="Swarm robotics"></div> *Try: raise cohesion alone and watch the flock collapse toward one point; balance all three weights and watch the order parameter climb toward one coherent heading; then drop in a predator.* Connects to: [[#Behavior-based control|Behavior-based control]] · [[#Mobile robots|Mobile robots]] · [[#Aerial robots|Aerial robots]] · [[#Braitenberg vehicles|Braitenberg vehicles]] ### Aerial robots *Main article: — stub, article not yet written (W4) · See also: [[Multirotor]], [[Unmanned_aerial_vehicle]]* > **Stub.** *Quadcopter* has no Wikitube article yet — and its sim is a true 3-D scene. Four independent rotor > thrusts sum to a net thrust and to roll, pitch and yaw torques from their differences; a PD attitude loop holds > the vehicle level in hover, tilting the target commands a translation, and a rotor-failure button shows how far > the remaining three can (or can't) compensate. The hexacopter variant extends the same thrust-allocation math to > six rotors, trading simplicity for redundancy. <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/robotics/Quadcopter.html" data-title="Quadcopter"></div> *Try: tilt the attitude target and watch the vehicle translate; then fail one rotor and see whether the PD loop can still hold attitude on three.* Connects to: [[#Actuators|Actuators]] · [[#The inverted pendulum|Inverted pendulum]] · [[#Sensors|Sensors]] · [[#Swarms|Swarms]] ### Self-driving cars <!-- SECTIONSIMS:BEGIN g34 2026-09-19 - microsims the articles linked here play; generated by _tools/generate/g34_portal_section_sims.py; do not hand-edit inside --> *Microsims from the articles this section links:* <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/engineering/Ackermann_steering_geometry.html" data-title="Ackermann steering geometry"></div> <!-- SECTIONSIMS:END --> *Main article: — stub, article not yet written (W4) · See also: [[Vehicular_automation]], [[Robot_control]]. Link: [[Ackermann_steering_geometry]] (Engineering ENG17)* > **Stub.** *Self-driving car* has no Wikitube article yet. Pure pursuit tracks a lane on the bicycle model by > aiming at a point a lookahead distance ahead on the path and steering toward it — a short lookahead cuts corners > tightly and can oscillate, a long one smooths the path but tracks it more loosely, and the sensor range cone > shows how far ahead the vehicle can even see to plan that far. The vehicular-automation variant reframes the same > control problem against the SAE's own levels of automation, from no assistance to full self-driving. <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/robotics/Self-driving_car.html" data-title="Self-driving car"></div> *Try: shorten the lookahead distance until the path tracking starts to oscillate; then lengthen it and watch the car cut corners smoothly instead.* Connects to: [[#Model predictive control|Model predictive control]] · [[#Mobile robots|Mobile robots]] · [[#Localization and mapping|Localization and mapping]] · [[PORTAL_Engineering#Steering geometry and handling|Steering geometry]] ## Part V — Industry, interaction and learning *Serves [[Manufacturing_engineering]], [[Biomedical_engineering]], [[Human–computer_interaction]] and [[Machine_learning]].* The last part is where robots work: the industrial arm's repeatable cycle, the grip that closes on a part, the force rendered back to a human hand in teleoperation, the policy a robot learns rather than is given, and the exoskeleton that puts robotics on a human body. ### Industrial robots *Main article: — stub, article not yet written (W4) · See also: [[Pick-and-place_machine]], [[Automation]], [[Unimate]]* > **Stub.** *Industrial robot* has no Wikitube article yet. A pick-and-place cycle drives each axis through a > trapezoidal velocity profile — accelerate at the limit, cruise, decelerate at the limit — under fixed velocity and > acceleration ceilings, and the cycle time read out is just the area under that profile; at the place point, a > repeatability cloud (per ISO 9283) shows the small scatter around the same nominal point that real repeated moves > actually produce, never landing on the exact same spot twice. <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/robotics/Industrial_robot.html" data-title="Industrial robot"></div> *Try: tighten the acceleration limit and watch cycle time grow; zoom into the place point and read the repeatability cloud's spread.* Connects to: [[#Serial manipulators and workspace|Serial manipulators]] · [[#Servomechanisms|Servomechanisms]] · [[#Omnidirectional wheels|Omnidirectional wheels]] · [[#Grippers and end effectors|Grippers]] · [[PORTAL_Engineering#Cams and followers|Cams]] · [[PORTAL_Engineering#Tolerances and fits|Tolerances and fits]] ### Grippers and end effectors *Main article: — stub, article not yet written (W4) · See also: none minted — `Prosthesis` and `Haptic_technology` belong to [[#Haptics and teleoperation|Haptics]] / [[#Wearable robots|Wearable robots]]* > **Stub.** *Robot end effector* has no Wikitube article yet. A parallel-jaw friction grasp has to supply enough > clamping force that friction alone can support the object's weight plus whatever acceleration it's carried > through, with a safety factor on top: F ≥ m(g + a)/(2μ)·SF. A suction cup sidesteps friction entirely — its > holding force is just a pressure difference times contact area, F = Δp·A — and the select swaps between the two > grasp physics on the same object. <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/robotics/Robot_end_effector.html" data-title="Robot end effector"></div> *Try: raise the carried acceleration and watch the required grip force climb past what friction alone can hold; switch to the suction cup and read the pressure-driven force instead.* Connects to: [[#Industrial robots|Industrial robots]] · [[#Haptics and teleoperation|Haptics]] · [[PORTAL_Engineering#Friction, bearings and lubrication|Friction]] ### Haptics and teleoperation *Main article: — stub, article not yet written (W4) · See also: [[Teleoperation]], [[Remote_manipulator]], [[Human–robot_interaction]]* > **Stub.** *Haptic technology* has no Wikitube article yet. Render a virtual wall through a haptic device and > whether it feels solid or buzzes depends on one stability condition, K·T < 2b, tying the wall's stiffness to the > control loop's sample rate and the device's own damping — push stiffness too high or the sample rate too low and > a "passive" wall starts to vibrate on its own. Add a communication delay, as any real teleoperated robot has, and > the same passivity condition gets harder to satisfy the longer that delay runs. <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/robotics/Haptic_technology.html" data-title="Haptic technology"></div> *Try: raise wall stiffness at a fixed sample rate until the passivity condition fails and the wall starts to buzz; then add a teleoperation delay and watch the same instability appear sooner.* Connects to: [[#Servomechanisms|Servomechanisms]] · [[#Grippers and end effectors|Grippers]] · [[#Wearable robots|Wearable robots]] ### Robot learning *Main article: — stub, article not yet written (W4) · See also: [[Q-learning]]. Link: [[Reinforcement_learning]] (exists, dense)* > **Stub.** *Robot learning* has no Wikitube article yet. Tabular Q-learning runs on a small grid world with a > cliff along one edge: learning rate, discount factor and exploration rate ε all on sliders, and episode by episode > the value arrows sharpen from random into a path that skirts the cliff rather than risking the shortest route > along its edge — a direct picture of the exploration/exploitation trade-off that Q-learning is famous for > illustrating on exactly this map. <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/robotics/Robot_learning.html" data-title="Robot learning"></div> *Try: run episodes until the value arrows converge on a safe path; then push ε high and watch the policy keep exploring risky cliff-edge routes even after it already knows better.* Connects to: [[#Behavior-based control|Behavior-based control]] · [[#Model predictive control|Model predictive control]] · [[#Motion planning|Motion planning]] ### Wearable robots *Main article: — stub, article not yet written (W4) · See also: [[Prosthesis]], [[Bionics]], [[Biomimetics]]. Link: [[Material_selection]] (Engineering ENG30)* > **Stub.** *Exoskeleton (human)* has no Wikitube article yet. A sagittal knee during a squat needs a joint torque > that the muscles alone must supply — or that an exoskeleton can share, on an assist-fraction slider that trades > muscle effort for machine effort directly, with a body-mass slider changing the baseline demand underneath it; an > illustrative metabolic-cost proxy reads out the net effect. The prosthesis variant contrasts a passive ankle > (energy only stored and returned, never added) against a powered one (energy actively added at push-off), the > same distinction that separates every assistive device on this page from a simple spring. <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/robotics/Exoskeleton_(human).html" data-title="Exoskeleton (human)"></div> *Try: raise the assist fraction and watch the muscle-supplied share of the knee torque fall; switch the prosthesis variant from passive to powered and read the metabolic proxy change.* Connects to: [[#Walking and the zero moment point|Zero moment point]] · [[#Grippers and end effectors|Grippers]] · [[#Haptics and teleoperation|Haptics]] · [[#Legged locomotion and gaits|Legged locomotion]] · [[PORTAL_Engineering#Design for additive manufacturing|Additive manufacturing]] · [[PORTAL_Engineering#Material selection|Material selection]] ## Crossover with Engineering The two spines were planned together and cross at fourteen points, written into the Connects-to lines above and in [[PORTAL_Engineering]]'s own sections: [[#Forward kinematics|forward kinematics]] and [[#Degrees of freedom and mobility|degrees of freedom]] share their mathematics with [[PORTAL_Engineering#The four-bar linkage|the four-bar linkage]]; [[#Actuators|actuators]] draw on [[PORTAL_Engineering#Simple machines and mechanical advantage|simple machines]], [[PORTAL_Engineering#Gears and gear trains|gears]] and [[PORTAL_Engineering#Hydraulics and pneumatics|hydraulics]]; [[#Servomechanisms|servo resonance]] is [[PORTAL_Engineering#Vibration isolation|vibration isolation]] wearing a control-loop hat; [[#Manipulator dynamics|manipulator dynamics]] carries the same inertia term as [[PORTAL_Engineering#Flywheels and inertia|flywheels]]; [[#Grippers and end effectors|grasping]] closes on the same friction that holds [[PORTAL_Engineering#Friction, bearings and lubrication|a block on an incline]]; [[#Mobile robots|the bicycle model]] is exactly [[PORTAL_Engineering#Vehicle suspension|vehicle suspension]]'s and [[PORTAL_Engineering#Steering geometry and handling|steering geometry]]'s plant; industrial-robot repeatability is what [[PORTAL_Engineering#Tolerances and fits|tolerances]] budget for; a planned path is [[PORTAL_Engineering#Machining and CNC|a CNC toolpath]] by another name; and [[#Wearable robots|wearable robots]] are built from the same [[PORTAL_Engineering#Material selection|material selection]] and [[PORTAL_Engineering#Design for additive manufacturing|additive-manufacturing]] choices as everything else on that spine. `Mechatronics` is the named bridge article both portals point to; `Electric_motor` (a live `/electronics/` sim) is linked from three sections on each side. ## Book shelf On the shelf now, cited by section above and in the full plan (`_registry/plans/ROBOTICS_SECTIONS.md`): **109** Jensen (mobility, gears, the DC-motor line, lead screws — shared with the Engineering spine), **022** Woolf (PID, Ziegler–Nichols, Routh), **034** Teunissen (recursive least squares, the mathematics the Kalman update *is*), **014** Dyer–Davis (sensors, accuracy and precision, accelerometers), **012** Fiore (PWM, microcontrollers), **001** Sayama (boids, agent-based models), **021** Hallauer and **073** Cline (second-order dynamics, Lagrangian mechanics). No dedicated robotics textbook is on the shelf yet; the manipulator mathematics (kinematics, Jacobians, dynamics) is marked *standard* to Craig and Lynch–Park's textbook forms rather than a page-cited source. Four open texts are proposed additions, each pending a license check before it gets a shelf row: *Modern Robotics* (Lynch and Park, free PDF, stated CC BY-NC-SA), *Introduction to Autonomous Robots* (Correll, Hayes, Heckman and Roncone, MIT Press open access), *Planning Algorithms* (LaValle, free online), and *Feedback Systems* (Åström and Murray, free PDF). ## Work list — what is left 1. **W4 — child articles.** Twenty-six stub sections above need Main articles at wt-article density, the largest single item on this portal's list; the four thin legacy pages (PID controller, Kalman filter, Model predictive control, Sensor) need append-only promotion to skeleton parity, not a rewrite. 2. **Flagship draft.** `_registry/plans/drafts/WT!Robotics_Compendium.draft.md` — thirty sections of flagship prose around these same sims plus the hero's own opening, for MTN to place as a `WT!` page. The existing `Robotics` article (a 2026-06 collision page) is not rewritten; it gets a back-link block and a `Microsims` pointer instead. 3. **W5 — generator and registry.** `g30_robotics_sims.py`, a copy of the Engineering set's `g29`, has not been written yet; no `<!-- ROBSIM:<id> -->` marker exists in any article because no robotics article exists yet. 4. **W6/W7 — deploy.** `deploy_robotics_set.sh` exists and is ready; running it once puts the 38 staged sims (Parts I–II, the hero, ROB18–19) live for the first time. The other 64 sims still need a staging pass before that script can reach them. 304 thumbnails and four short films (Humanoid robot, plus three more) are rendered and waiting on the same deploy pass. 5. **One item only MTN can clear.** The hero's own source files (`specs/sims/Humanoid_robot.json`/`.js`) are write-protected on the Mac; a pending install script (`Claude outputs/engrun/hero_pending/install_hero.sh`) is ready to apply once those two files are unlocked. The full detail, including every agent report and every library fix along the way, is in `_registry/plans/ENGINEERING_ROBOTICS_HANDOFF.md`. | # | Section | Main article | Sim | Variants (built) | |---|---|---|---|---| | ROB1 | Forward kinematics | **stub** | **staged** | 3/3 staged | | ROB2 | Inverse kinematics | **stub** | **staged** | 2/2 staged | | ROB3 | Degrees of freedom and mobility | **stub** | **staged** | 2/2 staged | | ROB4 | Serial manipulators and workspace | **stub** | **staged** | 3/3 staged | | ROB5 | Parallel manipulators | **stub** | **staged** (3-D) | 3/3 staged | | ROB6 | Actuators | **stub** | **staged** | 2/2 staged, 1 link-only | | ROB7 | Servomechanisms | **stub** | **staged** | 4/4 staged | | ROB8 | Manipulator dynamics | **stub** | **staged** | 1/1 staged, 1 link-only | | ROB9 | The inverted pendulum | **stub** | **staged** | 3/3 staged | | ROB10 | Walking and the zero moment point | **stub** | **staged** | 2/2 staged (incl. the Humanoid_robot hero) | | ROB28 | Legged locomotion and gaits | **stub** | **staged** | 3/3 staged | | ROB11 | PID control | [[PID_controller]] | **built** | 4/4 built | | ROB12 | State estimation and the Kalman filter | [[Kalman_filter]] | **built** | 2/2 built | | ROB13 | Behavior-based control | **stub** | **built** | 1/1 built | | ROB29 | Braitenberg vehicles | **stub** | **built** | — | | ROB14 | Model predictive control | [[Model_predictive_control]] | **built** | 2/2 built | | ROB15 | Sensors | [[Sensor]] | **built** | 4/4 built, 1 staged | | ROB16 | Mobile robots | **stub** | **built** | 3/3 built | | ROB17 | Omnidirectional wheels | **stub** | **built** | 1/1 built | | ROB18 | Motion planning | **stub** | **staged** | 5/5 built, 1 exists | | ROB19 | Localization and mapping | **stub** | **staged** | 3/3 built | | ROB30 | Monte Carlo localization | **stub** | **built** | 3/3 built | | ROB20 | Swarms | **stub** | **built** | 2/2 built | | ROB21 | Aerial robots | **stub** | **built** (3-D) | 2/2 built | | ROB22 | Self-driving cars | **stub** | **built** | 2/2 built | | ROB23 | Industrial robots | **stub** | **built** | 3/3 built | | ROB24 | Grippers and end effectors | **stub** | **built** | — | | ROB25 | Haptics and teleoperation | **stub** | **built** | 3/3 built | | ROB26 | Robot learning | **stub** | **built** | 1/1 built, 1 link-only | | ROB27 | Wearable robots | **stub** | **built** | 3/3 built | **Totals:** 30 sections · 31 roots built (incl. the hero) · ~74 See-also variants built · 4 Main articles written, 26 stubbed below · 38 sims staged and ready to deploy, 64 built and waiting for a staging pass, 0 live today. The full roadmap — child articles, the g30 generator, media, the deploy handoff MTN still has to run — is `_registry/plans/ENGINEERING_ROBOTICS_HANDOFF.md`; a same-day publish checklist is `_registry/plans/PUBLISH_NOW_2026-09-19.md`. --- *Created 2026-09-19 (wt-portal §6 pass) · sourced from `_registry/plans/ROBOTICS_SECTIONS.md`, the 2026-09-19 build ledger (`engrun_status_2026-09-18.tsv`, `_3d_deploy_stage/robotics/`) and `PUBLISH_NOW_2026-09-19.md` · sibling: [[PORTAL_Engineering]] · index: [[PORTAL_INDEX]]* --- *Repopulated 2026-09-19 · append-only · source: _tools/generate/g34_portal_section_sims.py@00a28cb2 (players of the linked articles, each URL 200-checked) · 53 added · 0 deletions*