A Novel Pet Trajectory Prediction Method for Intelligent Plant Cultivation Robot

Abstract

Intelligent plant cultivation robots will have certain working obstacles due to various moving objects, such as pets. This letter studies how to predict the trajectory of pets to avoid collisions. In previous intelligent plant cultivation robot systems, target-tracking technology was used to track pets and avoid collisions. However, collisions often occur even after tracking the trajectory of the pet because the trajectory of the pet's movement is difficult to determine. Therefore, the Informer algorithm is used to predict the trajectory of pets in this letter. It is presented that collisions can be avoided based on the predicted trajectory. Experimental results on the data set show that this algorithm has better accuracy than most common algorithms, so trajectory prediction based on this algorithm can make the robot avoid obstacles better. Finally, by verifying two data sets, the public data set and the self-made data set, it is determined that the performance of the proposed method is optimized by 3%–50% compared with other methods.