Long before ADAS got sophisticated enough to steer a car into a parking space on its own, two much simpler technologies quietly solved the everyday problem of judging distance to things a driver can’t see directly: the low bumper edge, the pillar right behind the rear window, the kerb hidden under the boot. Ultrasonic parking sensors and the reversing camera tackle that problem in two completely different ways, and understanding why both still exist side by side, rather than one simply replacing the other, comes down to what each one genuinely cannot do.
The sensor that doesn’t see anything — it listens
A parking sensor is, at its core, tiny sonar: a piezoelectric transducer embedded in the bumper emits a short ultrasonic pulse, typically around 40 kHz, well above what a human ear can hear, and then times how long it takes for that pulse’s echo to bounce back off whatever’s in front of it. Since sound travels through air at a known, fairly constant speed, that round-trip time converts directly into distance — no camera, no image processing, no lighting required at all, which is exactly why parking sensors work identically in pitch darkness, blinding sun glare or thick fog, conditions that genuinely confuse a camera-based system. The car’s control unit turns that raw distance into the beeping most drivers know well: slower beeps further away, a solid tone once you’re closing in on something the sensor’s threshold considers dangerously close.
Why sensors have blind spots of their own
Ultrasonic sensors are genuinely blind to anything outside their emission cone, and to anything that doesn’t reflect sound well — a thin metal pole, a low kerb edge, or a surface angled sharply away from the sensor can scatter the pulse instead of bouncing it straight back, giving a weak or entirely missing echo even though the object is real and close. That’s why bumpers typically carry four to six sensors rather than one or two: each covers a narrow cone, and only overlapping several of them gives reasonably complete coverage across the width of a bumper. It’s also why the classic ultrasonic sensor only ever tells you “something is roughly this far away in this general zone” — it has no way to say what that something actually is, or exactly where within its cone it sits. That single limitation is precisely the gap a camera is good at filling.
The camera: rich detail, no real sense of distance
A reversing camera, almost always mounted near the number plate or boot handle with a wide-angle or fisheye lens to maximise the field of view in a tight space, gives the driver exactly what ultrasonic sensors can’t: a genuine picture of what’s actually back there — a child, a shopping trolley, the specific shape of a bollard. What it doesn’t give for free is distance: a single camera has no inherent way to know whether an object is small and close or large and far away, which is why most systems overlay dynamic guidelines on the image, curved lines that shift in real time with the steering angle to show the car’s projected path, calculated from steering geometry rather than measured from the image itself. More advanced surround-view systems stitch feeds from four or more cameras — front, rear, and one in each door mirror — into a single virtual bird’s-eye view of the car from directly above, a genuinely difficult real-time image-processing problem of blending several distorted wide-angle feeds into one seamless, undistorted composite, updated dozens of times a second as the car moves.
Two blind technologies, one useful combination
Neither system alone is good enough for what modern automatic parking assist needs: distance sensors give solid, lighting-independent numbers but no context; cameras give rich context but weak numbers. Combining both — ultrasonic distance readings fused with camera-based object recognition — is exactly what lets a modern assisted-parking system judge both how big a gap actually is and whether what’s defining its edges is a harmless kerb or something the car genuinely needs to avoid, the same basic sensor-fusion logic used at a much larger scale by the perception stack behind higher levels of autonomous driving, just compressed down to the few centimetres and few seconds that matter when reversing into a tight space.
The EU research pushing ultrasonic sensing past one dimension
The core limitation of a conventional parking sensor — it measures a single distance number per cone, with no sense of exact position or shape — is exactly what the EU-funded TOPOSENS project (“3D Ultrasonic Sensors for Automotive – Democratizing 3D Vision Through Sound”, Horizon 2020, SME Instrument Phase 2, coordinated by Toposens GmbH, Germany, €2.45M EU contribution on a €3.49M total budget, 2020-2022) set out to fix. Rather than the single-microphone, single-echo design behind today’s sensors, Toposens built a sensor with multiple microphones and a patented algorithm that converts overlapping echoes into a genuine 3D point cloud — X, Y and Z coordinates for every reflecting surface in range, not just a distance figure — explicitly targeting close-range obstacle detection, protection of small or low objects near and under the car, and the higher levels of automated driving where a single distance number per zone simply isn’t enough information to act safely on.
Photo: © Car-Shooters