An Approach for Autonomous Feeding Robot Path Planning in Poultry Smart Farm

SIMPLE SUMMARY: In the poultry feeding process, a feeding robot instead of manual feed delivery can solve the problems of high labor demand and untimely feeding. However, if the feeding robot is not guaranteed to travel in an optimal path, it will cause large amounts of unnecessary energy consumption. In order to obtain the minimum energy consumption travel path of the feeding robot, the energy consumption of the feeding robot was taken as the optimization objective of the path planning in this study, and the minimum energy consumption travel path planning algorithm was designed. The experiment results show that the minimum energy consumption travel path could be obtained by the algorithms proposed in this study at the specified time, and that they have more computing power. The methods proposed in this study can reduce the production cost of the poultry smart farm to a certain extent and promote the development of poultry smart farms. ABSTRACT: In order to solve the problems of poor feeding environment, untimely feeding and high labor demand in poultry smart farms, the development of feeding robots is imminent, while the research on path planning algorithms is an important part of developing feeding robots. The energy consumption of the feeding robot is one of the important elements of concern in the process of path planning. In this study, the shortest path does not mean that the feeding robot consumes the least energy, because the total mass of the feeding robot keeps changing during the feeding process. It is necessary to find the most suitable path so that the feeding robot consumes the lowest amount of energy during the feeding process. A branch and bound algorithm to calculate the minimum energy consumption travel path for small-scale buckets lacking feed is proposed. The lower bound of the branch and bound on the energy consumption is obtained by the approach of preferred selection of the set of shortest edges combined with the sequence inequality, and the upper bound could be obtained based on Christofides’s Heuristic algorithm. A double-crossover operator genetic algorithm based on an upper bound on energy consumption for large-scale buckets lacking feed is proposed, and different crossover operations are performed according to the relationship between the fitness value and the upper bound of energy consumption in order to find a better path. The experiment results show that the approach proposed in this study is efficient; for small-scale buckets lacking feed, a branch and bound algorithm could calculate the minimum energy consumption path of 17 points in 300 s, and for large-scale buckets lacking feed, a double-crossover operator genetic algorithm based on an upper bound on energy consumption could calculate the minimum energy consumption travel path within 30 points in 60 s. The result is more accurate compared to the genetic algorithm with a single crossover operator.