Neural Networks (NN) have been the forefront of growth in recent years due to their variety, the opportunities they provide and most importantly their dynamic nature. A control system for catering robots for path planning is proposed with the help of neural networks as a comparative study. Various parameters such as training time, performance of the network, forecasted distance are considered after iterating to obtain the optimal dataset using Probabilistic Roadmap (PRM) algorithm. Approximately 36% improvement in forecasted distance was obtained using neural networks when compared to the traditional PRM algorithm. © 2018 The Authors. Published by Elsevier Ltd.