Fast detection of center of gravity position using neural comparator circuit based on artificial neuron

This paper proposes a method to detect the center of gravity position of an object in a short time. The method may be used for real-time detection of the center of gravity position for a walking robot, a running vehicle, etc. The detection circuit employs some neural comparators based on the behavior of information processing of a biological neuron. The circuit calculates the ratio of the two signals generated from load cell sensors, and detects the center of the gravity position between the load cell sensors located at the both end of a beam. The detection error of the location is within about 3% at full-scale ratio form the circuit experiments. The detection time is about 200ns from the input signal of sensors until the output of the detection signal on average. A parallel processing circuit using a layer of neural comparators enables fast detection. © 2020 The Institute of Electrical Engineers of Japan.