Intermediate Robotics & Automation interview questions

150 questions with answers (RAA-151 to RAA-300). Try each one aloud before reading the answer.

Robot Fundamentals & Classification

RAA-151 · What is the difference between repeatability and accuracy in robots?

Repeatability is how closely the robot returns to the same taught point; accuracy is how closely it reaches a commanded point computed from a model. Industrial robots are usually far more repeatable than accurate.

RAA-152 · Why are industrial robots usually taught by pendant rather than programmed offline with raw coordinates?

Teaching records joint positions the robot actually reached, so it relies only on repeatability. Offline coordinates rely on absolute accuracy, which suffers from kinematic model errors, deflection and thermal growth unless the robot is calibrated.

RAA-153 · How do you calculate mobility using the Grübler–Kutzbach formula?

For spatial mechanisms, M = 6(N − 1 − J) + Σfᵢ, where N is the number of links including ground, J the number of joints and fᵢ the DOF of joint i. For planar mechanisms, M = 3(N − 1 − J) + Σfᵢ.

RAA-154 · Using Grübler's formula, what is the mobility of a planar four-bar linkage?

N = 4, J = 4 revolute joints, each f = 1. M = 3(4 − 1 − 4) + 4 = −3 + 4 = 1 DOF.

RAA-155 · What is the joint configuration notation of common robots (e.g., RRR, PPP)?

Letters list joints from base outward: Cartesian is PPP, cylindrical RPP, spherical RRP, articulated RRR and SCARA RRP (with a wrist R).

RAA-156 · What is a redundant robot?

A robot with more DOF than the task requires, such as a 7-DOF arm for a 6-DOF pose. The extra DOF form a null space used to avoid obstacles, joint limits or singularities.

RAA-157 · Why does a SCARA robot need only 4 DOF for most assembly tasks?

Typical assembly needs X, Y, Z and rotation about the vertical axis. SCARA provides exactly these, making it faster and cheaper than a 6-axis arm.

RAA-158 · Why do parallel robots have higher stiffness than serial robots?

Loads are shared across several closed chains, and actuators are near the base. In serial arms, link deflections and joint compliance accumulate along the chain.

RAA-159 · What factors determine robot selection for an application?

Payload (including tool), reach, repeatability, speed and cycle time, number of axes, environment (cleanroom, washdown, explosion-proof), mounting, footprint, integration cost and safety.

RAA-160 · What is a robot's reach and how does it relate to payload?

Reach is the maximum distance from the base axis to the wrist centre. Longer reach increases the moment from the payload, so payload ratings fall with distance and offset from the flange.

RAA-161 · What is a payload diagram (wrist load diagram)?

A chart from the manufacturer showing the allowable payload versus the distance of its centre of gravity from the flange. Exceeding it overloads wrist motors and gears.

RAA-162 · What are the main components of an industrial robot system?

The manipulator (arm), controller (computation and drives), teach pendant, end effector, sensors and safety devices, plus programming software.

RAA-163 · What is the difference between an AGV and a fixed industrial robot in terms of workspace?

A fixed robot's workspace is set by its links and joint limits around its base. A mobile robot's workspace is the navigable floor area, limited by map, battery and safety zones.

Forward & Inverse Kinematics

RAA-164 · Why does forward kinematics have a unique solution while inverse kinematics does not?

FK is a direct evaluation of a function of the joint angles. IK inverts that nonlinear function, which can map several joint configurations (like elbow-up and elbow-down) to the same pose, or none if the pose is unreachable.

RAA-165 · What are Denavit–Hartenberg (DH) parameters?

Four parameters per link describing successive joint frames: θ (joint angle about z), d (link offset along z), a (link length along x) and α (link twist about x).

RAA-166 · Which DH parameter is the variable for revolute and prismatic joints?

For a revolute joint θᵢ is the variable and d, a, α are constant. For a prismatic joint dᵢ is the variable.

RAA-167 · How is the DH link transform constructed?

ⁱ⁻¹Tᵢ = Rot(z, θᵢ)·Trans(z, dᵢ)·Trans(x, aᵢ)·Rot(x, αᵢ) in the standard convention. The full FK is the product T = ⁰T₁·¹T₂·…·ⁿ⁻¹Tₙ.

RAA-168 · What is the difference between standard and modified DH conventions?

Standard (distal) DH attaches frame i at the far end of link i. Modified (proximal, Craig) DH attaches it at the near end and applies Rot(x, α) and Trans(x, a) before θ and d. Mixing them gives wrong results.

RAA-169 · Derive the forward kinematics of a planar 2-link arm.

x = L₁cosθ₁ + L₂cos(θ₁ + θ₂), y = L₁sinθ₁ + L₂sin(θ₁ + θ₂), where L₁, L₂ are link lengths and θ₁, θ₂ joint angles.

RAA-170 · For a planar 2R arm with L₁ = L₂ = 1 m, θ₁ = 90°, θ₂ = −90°, where is the end effector?

x = 1·cos90° + 1·cos0° = 1 m, y = 1·sin90° + 1·sin0° = 1 m. The tip is at (1, 1) m.

RAA-171 · How is the elbow angle found in the inverse kinematics of a planar 2R arm?

cosθ₂ = (x² + y² − L₁² − L₂²)/(2L₁L₂), and θ₂ = ±acos(...), giving elbow-down and elbow-up solutions. Then θ₁ = atan2(y, x) − atan2(L₂sinθ₂, L₁ + L₂cosθ₂).

RAA-172 · Why is atan2 used instead of atan in kinematics?

atan2(y, x) uses the signs of both arguments to return the angle in the correct quadrant over −π to π, and handles x = 0 without division by zero.

RAA-173 · What is the difference between analytical and numerical inverse kinematics?

Analytical (closed-form) IK gives exact solutions quickly and lists all branches, but exists only for certain geometries. Numerical IK iterates using the Jacobian, works for any arm but can be slow, find only one solution or fail to converge.

RAA-174 · How many IK solutions can a typical 6-axis industrial arm with spherical wrist have?

Up to 8 for a given pose: shoulder left/right × elbow up/down × wrist flip/no-flip.

RAA-175 · What is kinematic decoupling?

For arms with a spherical wrist, solving IK by first finding the wrist centre (pose minus tool offset along the approach axis) using the first three joints, then solving the last three joints for orientation.

RAA-176 · What is gimbal lock?

Loss of one rotational DOF in an Euler-angle representation when two rotation axes align, for example pitch = ±90° in roll-pitch-yaw, making the angles non-unique.

Robot Dynamics

RAA-177 · Why is the manipulator inertia matrix configuration dependent?

The effective inertia seen at a joint depends on how far the outer links' masses are from its axis, which changes as other joints move. A stretched arm has much higher shoulder inertia than a folded one.

RAA-178 · What is the holding torque for a single horizontal link of mass m and length L with its centre of mass at L/2?

τ = m·g·L/2. For m = 5 kg and L = 0.8 m, τ = 5 × 9.81 × 0.4 ≈ 19.6 N·m.

RAA-179 · How does a gear ratio N reflect load inertia to the motor?

Reflected inertia is J_load/N². With N = 100, a 10 kg·m² link appears as 0.001 kg·m² at the motor, which is why high gearing decouples joints.

RAA-180 · What is the optimal gear ratio for inertia matching?

N = √(J_load/J_motor), which maximises load acceleration for a given motor torque by making reflected load inertia equal motor inertia.

RAA-181 · What is gravity compensation?

Adding the computed gravity torque G(q) to the control command so the joint controller only has to handle errors and accelerations, improving accuracy and reducing integral wind-up.

RAA-182 · What is the effect of friction in robot joints?

Coulomb (constant), viscous (velocity-proportional) and Stribeck effects cause tracking errors, stick-slip at low speeds and limit-cycles. They are compensated with friction models or high-gain feedback.

RAA-183 · What is backlash and how does it affect robots?

Play between gear teeth that causes lost motion on direction reversal, reducing accuracy and potentially causing limit cycles. Harmonic drives and preloaded gears minimise it.

RAA-184 · What is a counterbalance in industrial robots?

A spring, gas strut or counterweight on a heavy axis, usually axis 2, that offsets gravity torque so smaller motors and gears can be used.

RAA-185 · What are the dynamic parameters that must be identified for model-based control?

For each link: mass, centre-of-mass position, the six independent inertia tensor components and friction coefficients. Often only base (identifiable) parameter combinations are estimated.

RAA-186 · What is the property Ṁ − 2C being skew-symmetric used for?

With proper choice of C, xᵀ(Ṁ − 2C)x = 0 for all x. It expresses energy conservation and is used in Lyapunov stability proofs of robot controllers.

RAA-187 · How do you compute the torque to accelerate a motor and load?

τ = (J_motor + J_load/N²)·α_motor + τ_friction + τ_load/N. Peak torque occurs during acceleration, which sizes the motor.

RAA-188 · What is a structural resonance in a robot arm and why does it matter?

A natural frequency of the flexible links, gears or base. If the control bandwidth or motion profile excites it, the arm vibrates, so bandwidth is kept below it or notch filters and input shaping are applied.

Trajectory Planning

RAA-189 · Why plan trajectories in joint space rather than Cartesian space?

Joint-space trajectories avoid IK at every step, cannot pass through singularities mid-motion, respect joint velocity limits easily and are faster. The TCP path is not straight, however.

RAA-190 · When is Cartesian-space trajectory planning required?

When the tool path matters, such as welding, gluing, cutting, inserting along an axis or moving close to obstacles in a straight line.

RAA-191 · What are the coefficients of a cubic trajectory from q₀ to q_f in time T with zero end velocities?

a₀ = q₀, a₁ = 0, a₂ = 3(q_f − q₀)/T², a₃ = −2(q_f − q₀)/T³.

RAA-192 · A joint moves 90° in 2 s with a rest-to-rest cubic. What is its peak velocity?

Peak velocity occurs at T/2 and equals 1.5·Δq/T = 1.5 × 90°/2 = 67.5°/s.

RAA-193 · Why use a quintic (5th-order) polynomial instead of a cubic?

A quintic has six coefficients, so it can match position, velocity and acceleration at both ends, giving continuous acceleration and avoiding infinite jerk at the start and end.

RAA-194 · For a trapezoidal profile with distance D, max velocity v and acceleration a, what is the move time?

If D ≥ v²/a, T = D/v + v/a. Otherwise the profile is triangular with T = 2√(D/a).

RAA-195 · A trapezoidal move: D = 1 m, v_max = 0.5 m/s, a = 1 m/s². What is the move time?

v²/a = 0.25 m ≤ 1 m, so it is trapezoidal. T = D/v + v/a = 2 + 0.5 = 2.5 s.

RAA-196 · What is trajectory time scaling?

Separating a path s ∈ [0, 1] from its timing s(t), so speed can be changed or limits enforced by changing s(t) only, without replanning the geometry.

RAA-197 · How are multiple joints synchronised in PTP motion?

The slowest joint, limited by its velocity and acceleration, sets the move time. The other joints are scaled to finish at the same time.

RAA-198 · What is spline interpolation in trajectory planning?

Fitting piecewise polynomials, such as cubic or B-splines, through multiple via points with continuous velocity and acceleration at the joins.

RAA-199 · Why must Cartesian orientation be interpolated with quaternions (slerp)?

Linear interpolation of Euler angles or matrix elements gives non-uniform or invalid rotations. Spherical linear interpolation (slerp) rotates at constant angular rate along the shortest arc.

RAA-200 · What is a time-optimal trajectory?

The fastest trajectory along a given path subject to joint velocity, acceleration and torque limits. It is typically bang-bang in acceleration, with at least one constraint active at every instant.

Robot Control

RAA-201 · What is independent joint control?

Treating each joint as a separate SISO system with its own PID loop and treating coupling and gravity as disturbances. It works well with high gear ratios.

RAA-202 · What is computed torque control?

Using the dynamic model to linearise the robot: τ = M(q)(q̈_d + K_d ė + Kₚe) + C(q, q̇)q̇ + G(q). With an exact model the error dynamics become linear and decoupled.

RAA-203 · What is feedforward control and why is it used in robots?

Adding a model-based command, such as inertia × desired acceleration plus gravity and friction, to the feedback output. It reduces tracking lag without raising feedback gains.

RAA-204 · What is integral windup and how is it prevented?

When an actuator saturates, the integral term keeps accumulating and causes large overshoot later. Prevent it with clamping, back-calculation or conditional integration.

RAA-205 · How does the derivative term amplify noise?

Differentiation boosts high-frequency components, so sensor noise is amplified. A low-pass filter on the D term, or taking the derivative of the measurement only, limits this.

RAA-206 · What is the difference between impedance and admittance control?

Impedance control measures motion and outputs force, suited to torque-controlled, backdrivable robots. Admittance control measures force and outputs motion commands, suited to stiff, position-controlled industrial robots with a F/T sensor.

RAA-207 · What is hybrid position/force control?

Splitting task space with a selection matrix into directions that are position-controlled and directions that are force-controlled, such as pressing down while sliding along a surface.

RAA-208 · What is the Ziegler–Nichols tuning method?

Increase Kₚ with I and D off until sustained oscillation at ultimate gain K_u and period T_u. Then for PID set Kₚ = 0.6K_u, Tᵢ = T_u/2, T_d = T_u/8.

RAA-209 · What is control bandwidth?

The frequency up to which the closed-loop system follows commands, usually the −3 dB point. Higher bandwidth gives faster response but is limited by resonances, noise and sampling rate.

RAA-210 · Why must the inner control loop be faster than the outer loop?

So the outer loop sees the inner loop as nearly ideal. A common rule is 5–10× separation in bandwidth, for example current loop at kHz rates and position loop at hundreds of Hz.

RAA-211 · What is visual servoing?

Using camera feedback directly in the control loop. Position-based visual servoing controls a 3D pose estimate; image-based visual servoing controls image feature errors.

RAA-212 · What is a Lyapunov function used for in robot control?

A positive-definite energy-like function whose derivative along system trajectories is negative (semi-)definite, proving stability without solving the equations of motion.

Sensors & Perception

RAA-213 · An incremental encoder has 2,500 lines with quadrature decoding. What is its resolution?

Quadrature gives 4 counts per line: 10,000 counts/rev, so resolution is 360°/10,000 = 0.036°.

RAA-214 · How does quadrature encoding determine direction?

Two channels, A and B, are 90° out of phase. Which channel leads tells the direction, and counting all edges gives 4× resolution.

RAA-215 · What are the trade-offs between motor-side and link-side encoders?

Motor-side encoders gain resolution from the gear ratio but cannot see gear backlash or flexibility. Link-side (output) encoders measure true joint position but have lower effective resolution and can complicate control.

RAA-216 · How does a time-of-flight distance sensor compute range?

d = c·t/2, where c is the speed of light (≈3 × 10⁸ m/s) and t the round-trip time. A 1 m range is about a 6.7 ns round trip.

RAA-217 · Why do IMU-based orientation estimates drift?

Gyroscope bias and noise are integrated over time, so angle errors grow without bound. Accelerometer (gravity) and magnetometer references correct roll, pitch and yaw drift.

RAA-218 · What is a complementary filter?

A filter that blends high-pass-filtered gyro integration with low-pass-filtered accelerometer angle, e.g. θ = α(θ + ω·Δt) + (1 − α)θ_acc, combining short-term and long-term accuracy.

RAA-219 · What is sensor fusion?

Combining data from multiple sensors, such as wheel odometry, IMU, GPS and LiDAR, to get estimates more accurate and robust than any single sensor provides.

RAA-220 · How does a Kalman filter work?

It recursively predicts the state with a motion model and its covariance, then corrects using measurements weighted by the Kalman gain, K = PHᵀ(HPHᵀ + R)⁻¹. It is optimal for linear systems with Gaussian noise.

RAA-221 · What is the difference between PNP and NPN sensor outputs?

A PNP (sourcing) output switches the load to +24 V. An NPN (sinking) output switches the load to 0 V. The PLC input card must match the type.

RAA-222 · What limits ultrasonic sensors?

Wide beam angle, specular reflections from angled smooth surfaces, absorption by soft materials, temperature-dependent speed of sound, a minimum blind zone and crosstalk between sensors.

RAA-223 · How does a strain-gauge force/torque sensor work?

Strain gauges bonded to an elastic structure change resistance under load. Wheatstone bridges convert this to voltage, and a calibration matrix maps the signals to Fx, Fy, Fz, Tx, Ty, Tz.

RAA-224 · What is the difference between 2D and 3D LiDAR?

A 2D LiDAR scans a single plane, adequate for floor navigation. A 3D LiDAR uses multiple beams or scanning to capture a volume, detecting overhangs and objects at different heights.

RAA-225 · What is sensor calibration and why is it needed?

Determining the mapping between raw readings and true values, including offset, gain and misalignment. Without it, systematic errors corrupt control and perception.

Computer Vision for Robotics

RAA-226 · What is machine vision used for in industrial automation?

Inspection (defects, presence, completeness), guidance (locating parts for pick-and-place), measurement (gauging dimensions) and identification (barcodes, 2D codes, OCR).

RAA-227 · What is the pinhole camera model?

A projection model in which a 3D point (X, Y, Z) maps to image coordinates u = f_x·X/Z + c_x, v = f_y·Y/Z + c_y, with focal lengths f_x, f_y and principal point (c_x, c_y).

RAA-228 · What are intrinsic and extrinsic camera parameters?

Intrinsics (focal lengths, principal point, skew, lens distortion) describe the camera itself. Extrinsics (rotation R and translation t) describe the camera's pose relative to the world.

RAA-229 · How is camera calibration performed?

By imaging a known target, such as a checkerboard, from multiple poses and solving for intrinsics and distortion that minimise reprojection error, for example with Zhang's method.

RAA-230 · What is lens distortion and how is it corrected?

Radial distortion (barrel or pincushion) and tangential distortion bend straight lines. A calibrated polynomial model is used to undistort images before measurement.

RAA-231 · How is depth computed from a stereo camera?

Z = f·B/d, where f is focal length in pixels, B the baseline and d the disparity in pixels. Depth error grows with Z², so accuracy drops with distance.

RAA-232 · A stereo pair has f = 700 px and B = 0.12 m. What is the depth at a disparity of 21 px?

Z = f·B/d = 700 × 0.12/21 = 4.0 m.

RAA-233 · What is hand-eye calibration?

Finding the fixed transform between the camera and the robot flange (eye-in-hand) or between the camera and the robot base (eye-to-hand), usually by solving AX = XB from multiple robot poses.

RAA-234 · What are image features such as SIFT and ORB?

Distinctive keypoints with descriptors that are robust to scale, rotation and lighting, used for matching, object recognition and visual odometry. ORB is a fast binary alternative to SIFT.

RAA-235 · What is the difference between eye-in-hand and eye-to-hand cameras?

Eye-in-hand cameras are mounted on the robot and move with the tool, giving close-up views. Eye-to-hand cameras are fixed in the cell and see the whole workspace but may be occluded by the robot.

RAA-236 · How are CNNs used in robot vision?

Convolutional neural networks learn features from data for classification, detection (e.g. YOLO), segmentation and grasp prediction, handling variation that rule-based vision struggles with.

RAA-237 · What is morphological processing?

Operations on binary images using a structuring element: erosion shrinks, dilation grows, opening removes small noise and closing fills small holes.

Path Planning & Navigation

RAA-238 · What makes an A* heuristic admissible and consistent?

Admissible means it never overestimates the true cost to goal. Consistent means h(n) ≤ c(n, n′) + h(n′) for every edge, which also implies admissibility.

RAA-239 · Why is Euclidean distance admissible on an 8-connected grid?

The straight line is the shortest possible route between two points, so it never exceeds the actual path cost through grid cells.

RAA-240 · How does PRM work?

Probabilistic Roadmap samples random collision-free configurations, connects nearby ones with collision-free local paths to form a graph, then searches the graph. It suits multiple queries in static environments.

RAA-241 · What is the difference between RRT and RRT*?

RRT finds a feasible path quickly but not an optimal one. RRT* rewires the tree to reduce cost as samples are added and is asymptotically optimal.

RAA-242 · What is the local minimum problem of potential fields?

The robot can get stuck where attractive and repulsive forces cancel, such as in a U-shaped obstacle, without reaching the goal.

RAA-243 · What is obstacle inflation in a costmap?

Expanding obstacles by the robot's radius plus a safety margin so the robot can be planned as a point while still clearing obstacles.

RAA-244 · What is the Dynamic Window Approach (DWA)?

A local planner that samples velocity commands (v, ω) reachable within the next control cycle, simulates short trajectories and scores them on goal heading, obstacle clearance and speed.

RAA-245 · What is the visibility graph method?

Connecting the start, goal and polygon obstacle vertices with all mutually visible straight segments. Searching this graph yields the shortest path in 2D, but it grazes obstacles.

RAA-246 · What is a Voronoi diagram used for in path planning?

It gives paths that maximise clearance from obstacles, trading path length for safety.

RAA-247 · Why are sampling-based planners preferred for robot arms?

Arm C-spaces are 6- or 7-dimensional, and grid-based search grows exponentially with dimension. Sampling-based planners only need collision checks and scale far better.

RAA-248 · What is the difference between 4-connected and 8-connected grids?

A 4-connected grid moves only along axes; an 8-connected grid also allows diagonals at cost √2, giving shorter and more natural paths.

RAA-249 · What is path smoothing?

Post-processing a jagged planned path, for example by shortcutting or fitting splines, to reduce length and curvature while remaining collision-free.

RAA-250 · What is D* (or D* Lite)?

An incremental search algorithm that efficiently repairs a path when edge costs change, such as newly detected obstacles, without replanning from scratch.

Mobile Robots & SLAM

RAA-251 · What are the kinematic equations of a differential-drive robot?

v = r(ω_R + ω_L)/2, ω = r(ω_R − ω_L)/L, with wheel radius r and track width L. Then ẋ = v·cosθ, ẏ = v·sinθ, θ̇ = ω.

RAA-252 · A differential-drive robot has r = 0.05 m, L = 0.4 m, ω_R = 12 rad/s, ω_L = 8 rad/s. Find v and ω.

v = 0.05 × (12 + 8)/2 = 0.5 m/s. ω = 0.05 × (12 − 8)/0.4 = 0.5 rad/s.

RAA-253 · What is particle-filter localization (Monte Carlo Localization)?

Representing the pose belief by weighted samples. Each particle is moved with the motion model, weighted by how well the sensor scan matches the map, then resampled.

RAA-254 · What is AMCL?

Adaptive Monte Carlo Localization: a particle filter that adapts the number of particles (KLD sampling), widely used for 2D LiDAR localisation in ROS.

RAA-255 · What is the difference between EKF-SLAM and graph-based SLAM?

EKF-SLAM keeps a joint Gaussian over pose and landmarks and updates it each step, scaling as O(n²). Graph SLAM stores poses and constraints as a graph and optimises them together, scaling to large maps.

RAA-256 · What is scan matching?

Aligning a new LiDAR scan with a previous scan or map, for example with ICP or correlative matching, to estimate relative motion.

RAA-257 · What is the ICP algorithm?

Iterative Closest Point: repeatedly match each point to its nearest neighbour in the other cloud and solve for the rigid transform minimising squared distances until convergence. It needs a good initial guess.

RAA-258 · What is visual odometry?

Estimating camera motion from successive images by tracking features. Monocular visual odometry has scale ambiguity; stereo or IMU fusion resolves scale.

RAA-259 · What is the kidnapped robot problem?

The robot is moved to an unknown position without being told. Localization must detect the failure and recover, which particle filters handle by injecting random particles.

RAA-260 · Why does wheel odometry drift?

Wheel slip, uneven floors, wheel radius and track width errors, and encoder quantisation cause errors that integrate over time, especially in heading.

RAA-261 · What is the difference between a topological and a metric map?

A metric map stores geometry with coordinates, such as an occupancy grid. A topological map stores places as nodes and their connectivity as edges.

RAA-262 · What is the minimum turning radius of an Ackermann vehicle with wheelbase L and max steering angle δ?

R = L/tanδ, measured to the rear axle centre. With L = 2.5 m and δ = 30°, R ≈ 2.5/0.577 ≈ 4.3 m.

ROS & ROS 2

RAA-263 · What is DDS and why does ROS 2 use it?

Data Distribution Service: an OMG publish/subscribe standard with automatic discovery and configurable QoS. It removes the need for a ROS master and supports reliable, real-time and multi-robot communication.

RAA-264 · What are ROS 2 QoS policies?

Settings such as reliability (reliable vs best effort), durability (volatile vs transient local), history depth, deadline and liveliness. Publisher and subscriber profiles must be compatible to connect.

RAA-265 · When should you use best-effort rather than reliable QoS?

For high-rate sensor streams like camera images or LiDAR where the latest sample matters more than every sample. Retransmitting stale data would only add latency.

RAA-266 · What is a ROS 2 lifecycle (managed) node?

A node with defined states (unconfigured, inactive, active, finalized) and transitions, letting a system manager configure and activate nodes in a controlled order.

RAA-267 · What are executors and callback groups in ROS 2?

Executors schedule node callbacks on one or more threads. Callback groups define whether callbacks may run concurrently (reentrant) or must be serialised (mutually exclusive).

RAA-268 · What is the standard frame chain map → odom → base_link?

map is a fixed world frame; odom is continuous but drifts; base_link is fixed to the robot. Localization publishes map→odom, and odometry publishes odom→base_link.

RAA-269 · What is a ROS bag?

A file that records topic messages with timestamps for later playback, debugging and dataset creation. ROS 2 uses ros2 bag with SQLite3 or MCAP storage.

RAA-270 · What is MoveIt?

The ROS motion-planning framework for manipulators, providing IK, collision checking, planners (OMPL and others), trajectory processing and controller interfaces.

RAA-271 · What is Nav2?

The ROS 2 navigation stack, providing localization, global and local planners, costmaps, recovery behaviours and behaviour-tree-based task orchestration for mobile robots.

RAA-272 · What is ros2_control?

A framework that abstracts robot hardware via hardware interfaces and runs controllers, such as joint trajectory or diff-drive controllers, in a real-time-capable control loop.

RAA-273 · What are ROS parameters?

Named configuration values, such as gains, frame names and rates. In ROS 2 they are owned by each node, declared with types and can be changed at runtime with callbacks.

RAA-274 · What is a ROS 2 DOMAIN_ID?

An integer that partitions DDS discovery so nodes only see others on the same domain, isolating multiple robots or teams on one network.

RAA-275 · What is the difference between colcon and catkin?

catkin is the ROS 1 build system based on CMake. colcon is the ROS 2 build tool that builds packages using ament_cmake, ament_python or other build types.

Industrial Automation (PLC, SCADA, Cells)

RAA-276 · What is PLC ladder logic?

A graphical language resembling relay circuits, with rungs of normally open and normally closed contacts, coils, timers and counters between two power rails.

RAA-277 · How is a start/stop seal-in (latching) circuit built in ladder logic?

A normally open Start contact in parallel with a contact of the Motor coil, in series with a normally closed Stop contact, driving the Motor coil. Once started, the motor holds itself on until Stop is pressed.

RAA-278 · Why should the physical stop button be wired normally closed?

A broken wire or loose terminal then reads as a stop, so the system fails safe. A normally open stop would silently stop working if the wire broke.

RAA-279 · What is the difference between TON, TOF and TP timers?

TON delays turning on: output goes true after input stays true for the preset. TOF delays turning off: output stays true for the preset after input goes false. TP gives a fixed-length pulse.

RAA-280 · What is the difference between a CTU and a CTD counter?

CTU counts up on each rising edge and sets its output when the count reaches the preset. CTD counts down from the preset and sets its output at zero.

RAA-281 · What is Sequential Function Chart (SFC) used for?

Programming sequential processes as steps and transitions, such as machine cycles. It makes states and conditions visible and simplifies troubleshooting.

RAA-282 · How is a 4–20 mA signal scaled to engineering units?

Value = (I − 4)/(16) × (Max − Min) + Min. For a 0–10 bar transmitter, 12 mA gives (12 − 4)/16 × 10 = 5 bar.

RAA-283 · Why is 4–20 mA preferred over 0–10 V for field signals?

Current is unaffected by wire resistance and less noise-sensitive over long runs, and the 4 mA live zero lets a broken wire (0 mA) be detected.

RAA-284 · What is OPC UA?

A platform-independent industrial communication standard with an information model, client/server and pub/sub transports and built-in security, used to link PLCs, SCADA and MES.

RAA-285 · What is the automation pyramid (ISA-95 levels)?

From bottom: field devices (level 0), control such as PLCs (1), supervisory SCADA/HMI (2), manufacturing operations MES (3) and business planning ERP (4).

RAA-286 · How does a robot controller interface with a cell PLC?

Usually via fieldbus (PROFINET, EtherNet/IP, EtherCAT) or digital I/O, exchanging handshake signals such as program select, start, cycle complete, in-zone and fault, while the PLC sequences the cell.

RAA-287 · What is OEE?

Overall Equipment Effectiveness = Availability × Performance × Quality. For example 0.90 × 0.95 × 0.99 ≈ 84.6 %.

RAA-288 · What is a PLC watchdog timer?

A timer that faults the CPU and puts outputs in a safe state if the program scan exceeds a set time, catching infinite loops or hangs.

Collaborative Robots & Safety

RAA-289 · How do cobots detect collisions?

By monitoring motor currents or joint torque sensors against the dynamic model. A deviation beyond a threshold triggers a stop or a reflex retraction.

RAA-290 · Why do cobots typically have lower payload and speed than industrial robots?

Limiting force and energy on contact requires low moving mass, limited speed and backdrivable or torque-sensed joints, which restricts payload and cycle speed.

RAA-291 · What is the difference between a collaborative robot and a collaborative application?

Safety comes from the application: tool, part, speed, forces and layout. A cobot holding a knife or a sharp part is not safe just because the arm is collaborative.

RAA-292 · What are the four collaborative operation modes?

Safety-rated monitored stop, hand guiding, speed and separation monitoring, and power and force limiting.

RAA-293 · What is the difference between transient and quasi-static contact?

Transient contact is a brief impact where the body part can recoil. Quasi-static contact is clamping or pinning of a body part between robot and an object, which has lower allowed limits.

RAA-294 · What is the difference between Category 0, 1 and 2 stops?

Stop Category 0 removes power immediately (uncontrolled). Category 1 brakes under control then removes power. Category 2 is a controlled stop with power kept on.

RAA-295 · What are Performance Levels (PL) in machine safety?

ISO 13849-1 ratings PL a to PL e of a safety function's ability to perform under foreseeable conditions. PL e is the highest; robot safety functions commonly target PL d.

RAA-296 · What is safety-rated reduced speed?

A monitored speed limit, often 250 mm/s at the TCP for manual (T1) teaching mode, enforced by the safety controller.

RAA-297 · What are safety zones and safe tool orientation in a robot controller?

Safety-rated software limits, such as joint ranges, Cartesian zones and tool orientation limits, monitored by a safety controller so the robot cannot enter forbidden areas even if programmed to.

RAA-298 · What is muting in light curtains?

Temporarily bypassing the curtain, using sensors, to let material pass on a conveyor while still detecting a person entering.

RAA-299 · What is a safety mat?

A pressure-sensitive floor mat that sends a stop or presence signal when someone stands on it, used to detect people in a hazard zone.

RAA-300 · Why must an E-stop be hardwired or safety-rated rather than handled in normal PLC logic?

Standard PLC logic and software can fail undetected. Safety functions need redundant, monitored channels (safety relay or safety PLC) to reach the required Performance Level.

Other levels: Basic · Advanced · Tricky · All Robotics & Automation questions