Machines can have more than five senses.
Human language encourages us to think in terms of sight, hearing, touch, smell and taste. Machines are less constrained. They can receive visible and infrared light, depth, pressure, acceleration, vibration, radio, temperature, chemical composition, electrical activity and many other forms of signal.
Machine olfaction is a useful example. Smell is not an image to classify. It involves chemical signatures, mixtures, concentration, drift and context. Turning those patterns into useful information requires sensing hardware, calibration and models that can reason under uncertainty.
One sensor rarely explains a situation.
A camera can see posture but not necessarily pressure. A microphone can hear a voice but not know whether a person is moving away. A wearable may detect physiological change without knowing its cause. Multimodal sensing combines partial views of the same reality.
Human signals are physical signals too.
Voice cadence, gaze, motion, facial behavior, skin temperature, heart-rate patterns and spatial behavior can all become inputs. But once signals concern people, the technical problem becomes inseparable from context, consent and uncertainty.
The interesting boundary is where sensing becomes perception.
A sensor measures. A sensing system selects, synchronizes, combines and qualifies signals. The next layer is to build a coherent model of what may be happening.