Grasshoppers, like most animals, try to avoid rapidly approaching objects; since it is very possible that they are a predator or some other danger to their physical integrity. In its nervous system there are neurons in charge of detecting threats and signaling when it is time to escape.
The LGMD ( lobula giant movement detector ) is a type of neuron that responds selectively to objects that approach the grasshopper’s eye. Numerous scientific studies have tried to translate its operation into an algorithm , whose implementation would be able to emulate with biological plausibility the way in which these neurons act to differentiate the different visual stimuli and react to collisions.
Take advantage of the “noise” of the nervous system
A common feature of LGMDs is that they receive information from the entire visual field . The process of integration of these signals is carried out by the dendritic tree of the neuron, the branches from which these cells of the nervous system receive information. In this process, noise, understood as unwanted modifications of the original signal, is ubiquitous and unavoidable. However, the reliability with which organisms carry out their behavior shows that the nervous system is designed in such a way that it is capable of handling this noise.
Our team, made up of researchers from the Institute for Research on Brain, Cognition and Behavior of the University of Barcelona (IR3C), has developed a computational model in which it is assumed that each of the input signals is contaminated by noise and that it has a response threshold that must be exceeded to provoke a nervous reaction. An adequate amount of noise can help otherwise undetectable signals exceed the detection threshold and initiate neuron response. In this way, the noise would be used constructively for the collision sensitive response calculation.
A model to improve automatic driving
This model, published in the journal PLOS Computional Biology , has been applied to the development of an algorithm that detects approaching objects and distinguishes them from other objects that enter the visual field. The TARGET algorithm, which would be like the retina of the grasshopper , transforms sequences of images into optical variables such as the angular size of an object projected on the observer’s retina and the rate of change of the angular size, that is, the speed with which it moves the object by the retina. This algorithm allows detecting moving objects from images and videos, and it is a tool that in the future could be used in the automatic driving of cars or airplanes.
Matthias Keil, researcher at the Institute for Research in Brain, Cognition and Behavior (IR3C) of the University of Barcelona. This article has been written in collaboration with the Communication Unit of the University of Barcelona.
