A Time-of-Flight Camera That Holds Up Outdoors
IDS has launched Nion, a 1.2-megapixel industrial time-of-flight camera running 30 fps over a 0.3 to 7.5 metre range in an IP67 housing with PoE, built for stray-light resistance and low motion blur.
IDS Imaging Development Systems has launched Nion, a 1.2-megapixel industrial time-of-flight camera delivering 30 frames per second with temporally stable depth data over a 0.3 to 7.5 metre measuring range. It is built on the onsemi AR0130 depth sensor, ships in an IP67 housing with Power over Ethernet, and is specified for minimal motion blur during rapid object movement and strong resistance to stray light for consistent indoor and outdoor results. IDS claims superior resolution, reduced noise and improved ambient light stability against standard time-of-flight cameras. Target applications are automation, robotics and logistics; Patrick Schick, product manager for 3D vision and imaging software, is quoted.
Stray-light resistance is the specification that decides whether a time-of-flight camera is usable in a real facility, and it is the one most often glossed over. A ToF camera works by emitting modulated infrared light and measuring the phase shift of the returning signal. Any other infrared source in the scene — sunlight through a roller shutter door, a halogen lamp, a heater, another ToF camera pointed the wrong way — adds energy that the sensor cannot distinguish from its own illumination, and the result is depth noise or outright invalid pixels. This is why so many depth-sensing pilots work in a lab and fail in a loading bay, and why the honest comparison between cameras in this class is made at the worst-lit point of the intended installation rather than on a datasheet.
The "temporally stable" phrasing is the other claim worth reading carefully, because stability matters more than raw accuracy for most applications. A depth value that is accurate on average but fluctuates frame to frame produces a point cloud that flickers, which breaks object tracking, makes plane fitting unreliable and forces heavy temporal filtering that in turn adds latency. For a robot deciding where to place a gripper, a slightly biased but steady measurement is far more useful than an unbiased noisy one — the bias can be calibrated out, the noise cannot.
The range and resolution place it clearly. Between 0.3 and 7.5 metres at 1.2 megapixels is the geometry of palletising, depalletising, bin picking at the near end, autonomous mobile robot navigation and obstacle detection, and volumetric measurement of parcels — not long-range outdoor perception and not precision metrology. For anyone evaluating, the practical test list is short and specific: the shiniest and darkest objects in the real material mix, the scene lit as it will actually be lit, an object moving at production speed, and a check of what the camera reports at the edges of its range, because ToF cameras tend to fail by returning a confident wrong number rather than by reporting nothing.