Multi-pruning-robot and multi-fertilizing-drone collaborative task assignment using multi-class teaching-learning-based optimization

Abstract

The development of intelligent agricultural robots and drones has greatly advanced smart agriculture. This paper investigates a multi-pruning-robot and multi-fertilizing-drone task assignment problem (MRMDTA) in a smart orchard, aiming to minimize the makespan of the whole robot-drone system. To address this issue, a mathematical model is formulated, and an innovative multi-class teaching-learning-based optimization (MTLBO) algorithm is proposed. The MTLBO algorithm integrates a multi-class teaching approach, where each class is led by both a teacher and a teaching assistant, enhancing learning efficiency across diverse groups. The algorithm operates through a well-structured, six-stage optimization process. Firstly, in the initialization stage, two heuristics based on greedy insertion are introduced. Subsequently, in the class division stage, a teacher and a teaching assistant are assigned to each class. Then, in the training stage, five heuristic search operators are designed. Following this, in the learning stage, a recombine crossover operator is presented for students to learn from teachers. Next, in the collaboration stage, a temporary class is formed. Finally, in the graduation stage, individuals with little search potential are eliminated. Extensive experimental results demonstrate that the proposed MTLBO algorithm outperforms state-of-the-art algorithms in terms of efficiency and solution quality.