A robot arm uses an absolute encoder on each joint. After the robot is powered off and moved manually, why can the controller still determine the correct joint positions when power is restored — without needing to re-home the arm?

The encoder stores position in non-volatile memory between power cycles An absolute encoder outputs a unique code for every angular position within a full revolution, so the joint angle is known immediately at power-on The robot controller estimates position from the last known motor current Absolute encoders send a continuous electrical signal that cannot be interrupted

A self-driving car uses both LIDAR and radar. In a heavy rainstorm, the LIDAR point cloud becomes noisy and unreliable. Why does radar remain effective in this situation?

Radar has a longer range than LIDAR so it can see around the rain Radar uses longer-wavelength radio waves that pass through rain droplets with minimal scattering, unlike the short-wavelength laser pulses used by LIDAR Radar measures temperature, not distance, so rain does not affect it Radar is mounted higher on the vehicle, above the rain

A stereo camera system has a baseline (distance between the two cameras) of 12 cm. An engineer wants to use it to reliably estimate the depth of objects up to 10 m away. What is the fundamental limitation the engineer must consider?

Stereo cameras can only measure depth in bright daylight conditions Stereo cameras require the two images to be taken simultaneously, which is mechanically impossible Depth accuracy degrades rapidly with distance because the pixel disparity between the two images becomes very small for distant objects, approaching the sub-pixel noise floor Stereo cameras cannot work if any part of the scene has texture

A mobile robot relies solely on its IMU (accelerometer + gyroscope) to track its position while navigating a large warehouse. After 10 minutes, the robot's estimated position is significantly wrong. What is the fundamental cause of this error?

IMUs only work when the robot is moving faster than 1 m/s The warehouse metal racking interferes with the IMU's magnetic compass Small constant biases and noise in the IMU sensors cause integration errors that accumulate over time, leading to position drift that grows without bound IMUs measure velocity, not acceleration, so they cannot track position at all

A robot arm is assembling a precision connector. A force/torque sensor at the wrist reads a sudden spike of 15 N in the downward direction just before the connector fully seats. What should the controller infer from this reading, and what should it do?

The connector is fully seated — the spike confirms contact and the robot should stop The connector has met resistance — likely misaligned — and the controller should reduce insertion force and perform a search motion to find the correct alignment The sensor is faulty and the robot should ignore the reading and continue at full speed The force spike means the assembly is complete and no further action is needed

An ultrasonic parking sensor and a LIDAR unit are both used for short-range obstacle detection on a robot. The ultrasonic sensor has a range of 4 m and accuracy of ±1 cm. Why would an engineer still choose LIDAR for tasks requiring precise object shape recognition?

LIDAR has longer range than ultrasonic sensors in all conditions LIDAR produces a dense point cloud that captures the 3D shape and edges of objects; ultrasonic sensors only return a single distance measurement along a wide cone beam Ultrasonic sensors cannot detect solid objects, only liquids LIDAR is cheaper than ultrasonic sensors for mass-market robots

A search-and-rescue robot is deployed inside a collapsed building where GPS signals cannot penetrate. Which combination of sensors would best allow the robot to estimate its position inside the structure?

GPS + camera — GPS gives global position and the camera fills in gaps IMU alone — it measures acceleration directly so no other sensor is needed IMU (for short-term dead reckoning) combined with LIDAR (for SLAM-based mapping and loop closure to correct drift) Ultrasonic sensors only — cheapest and most reliable in dusty environments

An RGB-D camera (like a Microsoft Azure Kinect) projects an infrared structured-light pattern onto a scene to measure depth. Why does this approach fail outdoors in bright sunlight?

Sunlight causes the RGB camera to overexpose, so depth cannot be computed from colour Structured-light cameras only work at temperatures below 25°C Bright sunlight contains strong infrared radiation that overwhelms the projected structured-light pattern, making the depth sensor unable to detect its own pattern The RGB-D camera requires a calibration target that cannot be used outdoors

A Kalman filter is used to fuse GPS (noisy position, ~5 m accuracy) with wheel odometry (precise short-term, but drifts) on a delivery robot. What does the Kalman filter do with these two data streams?

It always uses GPS and completely ignores odometry as it is less accurate It averages the GPS and odometry positions with equal weight at every time step It weights each sensor's contribution by its estimated uncertainty — relying more on odometry for smooth short-term tracking but using GPS to bound long-term drift It switches between GPS and odometry based on which one was updated most recently

RTK GPS (Real-Time Kinematic) achieves centimetre-level positioning accuracy compared to standard GPS (±5 m). How does RTK achieve this dramatic improvement?

RTK uses more satellites than standard GPS, improving triangulation accuracy RTK uses a higher-frequency signal that is less susceptible to interference A fixed base station with a known position broadcasts correction signals that account for atmospheric delays and satellite orbit errors, which the rover applies to its own measurements RTK measures the exact speed of light in real time to correct timing errors

A monocular camera mounted on a drone is used to estimate the 3D position of objects on the ground. What fundamental geometric limitation must the software overcome?

A monocular camera cannot detect objects smaller than 10 cm in diameter A single camera image collapses 3D space onto a 2D plane — depth information is lost and must be recovered indirectly through motion parallax or object size assumptions Monocular cameras cannot distinguish between colours in low-light conditions A single camera can only capture images at 30 frames per second, too slow for robotics

An industrial resolver (angle sensor) is specified for a robot joint in a steel mill where temperatures reach 150°C and strong electromagnetic interference is present. Why is a resolver preferred over an optical encoder in this environment?

Resolvers have higher angular resolution than optical encoders Resolvers use a rugged inductive (transformer) principle with no optical components or digital electronics in the sensor head, making them extremely resistant to heat, vibration, and electromagnetic interference Resolvers are cheaper to manufacture than optical encoders Resolvers can be used without any signal conditioning electronics