Fruit picking sequence planning method based on visual selection attention mechanism

Abstract

The invention discloses a fruit picking sequence planning method based on a visual selection attention mechanism. The method comprises the steps that a fruit image is collected and registered, fruit visual saliency is calculated, a fruit target is recognized and segmented, fruit picking decision attribute is analyzed, then the average saliency, the central depth value and the shielding coefficientof each fruit are taken as three decision elements for picking priority ranking, the decision element attribute value of each fruit is comprehensively evaluated by adopting a fruit picking priority ranking method, and the picking priority sequence of each fruit in the current scene is determined; and according to the picking priority sequence of each fruit, the three-dimensional coordinates of each fruit are sequentially sent to a fruit robot, and the fruit robot is driven to perform continuous picking. The fruit picking sequence planning method can pick the higher-quality fruits under the condition that energy consumption is relatively low, and has practical significance for promoting production and income increase of the fruit industry in China and intellectualization of automatic picking equipment.

Description

technical field [0001] The invention belongs to the field of visual bionics of agricultural robots, and in particular relates to a fruit picking sequence planning method based on a visual selection and attention mechanism. Background technique [0002] With the start of smart agriculture, there are more and more applications of smart agricultural systems. Fruit picking robots are an emerging field in smart agriculture. How to improve the efficiency and quality of fruit picking has always been the focus of research in this field. At present, there are two main picking methods for vision-based fruit picking robots: (1) Use the camera to continuously take pictures from multiple random directions until the fruit target is found, and then control the robot to pick; The uncertainty of the environment wastes a lot of repetitive actions, resulting in low efficiency. (2) Use the camera to take images at a distance, detect all fruit targets in the current scene at one time, and then ...

AI Technical Summary

Problems solved by technology

At present, there are two main picking methods for vision-based fruit picking robots: (1) Use the camera to continuously take pictures from multiple random directions until the fruit target is found, and then control the robot to pick; The uncertainty of the environment wastes a lot of repetitive actions, resulting in low efficiency

(2) Use the camera to take images at a distance, detect all fruit targets in the current scene at one time, and then control the robot to pick them one by one; but this approach creates a problem, that is, how to determine which fruit the robot picks first and which fruit

To solve this problem, there are mainly two picking methods. One is to pick from left to right and from top to bottom, but the distance between the fruit and the robot is not considered, so the efficiency cannot be guaranteed; the second is to plan the shortest path of the entire picking process. Go picking to reduce the consumption of the robot and improve the picking efficiency of the robot

However, the above fruit picking methods all ignore the quality of the picked fruit, resulting in easy picking of fruits with uneven quality, which increases the heavy burden of subsequent fruit quality sorting work, and the labor cost remains high, which directly affects fruit farmers. overall revenue

Images