- Details
-
Published on Monday, 26 February 2018 23:52
Random Position Object Sensing refers to applications where the goal is to either detect the presence of an “object”in an open delimited space or count the number of objects travelling through a constrained space, usually an enclosure.Examples of such applications are safety curtains, pill counters, seeds counter, and other objects dispensers. Many factors have to be taken into account for the system design, combining spatial, optical and electrical considerations. Just to name a few:
- Object optical properties (reflectivity)
- Enclosure optical properties
Hereafter are a few conceptual drawings showing the many possibilities:
1- 1 single LED / 1 row of detectors
This configuration is well suited for detecting and counting relatively large objects. The rule of thumb is that the object must be larger in its smallest dimension than the distance “d”. It may produce a count error if several objects transit more or less simultaneously in the enclosure.
The LED must be positioned and driven such that when the enclosure is empty, all detectors are “ON” (high logic state). Its emission pattern must be wide enough to illuminate the detectors located at each side, but is not limited.
2- 1 row of LEDs / 1 row of detectors
This configuration will be preferred to the previous one for applications where the enclosure or safety space is large and the size of the object very variable. This is typically the case for safety light curtains applications.
The LEDs must have a narrow emission pattern so that they illuminate only one detector. The test for that condition is that when LED(i) is OFF and all others are ON, detector(i) must be OFF (low logic state).
3- 2 dimensions matrix detector plane
In order to improve the capability of the system to detect small objects, one possibility is to place the detectors as shown on Fig.3: the minimum size of the object can be divided by two compared to Fig.1, and even more by increasing the numbers of rows and reducing the alignment shift from row to row.
All those concepts are helped by two characteristics of the detector LIM158FS:
- Its small size, that allows to create a fine mesh of detectors. The benefits are the possibility to detect small objects and the prevention of misses due to the absence of blind spots .
- The logic output function that allows to combine the multiple outputs in a single output by connecting them all together in a logic “AND”, without additional circuitry. See the schematic Fig.4.
Another benefit of the logic output is that each individual output has the same weight in the determination of the combined output, regardless of the amount of light received and the sensitivity of the detector. Therefore, the designer has more leeway in the choice and settings of the LEDs and detectors. For instance there is no need for narrow distributions or pairing emitters and detectors.
Beyond the concepts described above, other techniques may be utilized to enhance the performance of the system (for instance, less statistical misses, less false detections or better count accuracy), or add functionality like a mapping of the areas where detections occur.
For example, in a system including multiple LEDs, one can consider to energize the LEDs sequentially one by one, recording the response of the detectors, allowing to determine which LED triggered a detection. In the case of an object counter, it would separate the responses of two objects present simultaneously in the detection space.
Using AI and Deep Learning algorithms, it is possible to design a standard versatile system, able to accommodate various sizes and characteristics of the objects: for a given application, the system can learn to recognize the signature of that object and optimize the settings to capture this signature efficiently. The spatial / optical hardware design can be standard (i.e. not specific to one application), which results in lower manufacturing costs and faster time to market.
Download PDF