Automated arthropod detection system

Abstract

The invention is a system and method for searching a subject or substrate for arthropods and communicating information about any arthropods to a user. A preferred embodiment of an automated arthropod detection system includes an imaging subsystem, a processor, and a communication unit. Embodiments of an automated arthropod detection system may implemented in various configurations, which include, but are not limited to: a non-imaging configuration, a handheld configuration, a self-supporting configuration, and a wearable configuration. The method of the invention comprises using an embodiment of an automated arthropod detection system to obtain one or more digital images, process the one or more digital images using an arthropod recognition machine learning algorithm, and communicate the results from the arthropod recognition machine learning algorithm to the user of the system. The method of obtaining one or more digital images may involve accepting one or more images from a user or capturing one more more images using an image sensor.

I claim: 
     
         1 . A method of arthropod detection using an automated arthropod detection system, the method comprising:
 generating a sequence of movements to be executed at a plurality of future points in time by a powered arm,   executing the sequence of movements at the plurality of future points in time while also capturing a first one or more digital images of a subject or substrate from one or more positions using one or more image sensors mounted on the powered arm,   processing the first one or more digital images of the subject or substrate using an arthropod recognition machine learning algorithm, and   communicating the results from the arthropod recognition machine learning algorithm to a user of the system;   
       and wherein:
 the step of generating the sequence of movements to be executed at the plurality of future points in time by the powered arm comprises:
 capturing a second one or more digital images of the subject or substrate using the one or more image sensors mounted on the powered arm, and processing the second one or more digital images of the subject or substrate using a movement planning machine learning algorithm; 
 
 said movement planning machine learning algorithm is a machine learning algorithm that accepts the second one or more digital images of the subject or substrate as inputs and produces outputs that represent the sequence of movements to be executed at the plurality of future points in time by the powered arm, wherein said sequence of movements will allow the one or more image sensor sensors mounted on the powered arm to capture the first one or more digital images of the subject or substrate in such a way that said first one or more digital images of the subject or substrate include a plurality of the subject or substrate's surfaces; 
 the movement planning machine learning algorithm is one of: a supervised learning method, an unsupervised learning method, a semi-supervised learning method, a reinforcement learning method, an optimization technique, a convolutional neural network, a recurrent neural network, a deep learning model, or a generative-adversarial network; 
 the parameters of the movement planning machine learning algorithm are chosen by training on one or more of first labeled or first unlabeled data; 
 the sequence of movements represented by the output of the movement planning machine learning algorithm will additionally cause the powered arm to exert force on and reposition one or more parts of the subject or substrate, or one or more parts of an object that is near the subject or substrate; and 
 said sequence of movements will additionally allow the one or more image sensors mounted on the powered arm to capture the first one or more digital images of the subject or substrate in such a way that said first one or more digital images of the subject or substrate include surfaces of the subject or substrate that were not previously visible.