sensor noise

Sensor noise is the random, unwanted jitter that contaminates every real measurement. If you hold a perfectly still robot and read its distance sensor a hundred times, you will not get the same number a hundred times; you will get a cloud of slightly different values scattered around the true one. That scatter is the noise. It comes from unavoidable physics, such as the thermal rattling of electrons inside the circuit, tiny vibrations, electrical interference from nearby motors, and the act of rounding a smooth real-world value into discrete digital steps. Noise is different from a steady error: noise wobbles unpredictably from one reading to the next, whereas a fixed offset (called bias) leans every reading the same way.

Engineers usually describe noise by its size, often as a standard deviation, which says how widely the readings spread, and they assume it averages out to zero over time. That assumption is powerful, because it means you can fight noise by averaging many readings or by smoothing the signal, trading a little speed for a steadier number. Why does it matter? A robot that believed every noisy spike would twitch, brake, or steer at phantom obstacles. Knowing the noise level lets the robot weigh its sensors sensibly, leaning on a quiet, trustworthy one and discounting a jittery one. This idea of attaching an honest uncertainty to each measurement is the foundation of sensor fusion and the filters that keep robots steady.