Artificial Neural Networks (ANNs) are well known for oscillatory output when in a closed loop with the feedback from the system it is controlling. Here we propose a new neural controller for locomotion in linear modular robotic configurations, based on oscillatory output from simple ANNs. We investigate two different methods for controlling the action of each module based on the neural output, and also evolve neural controllers which have the ability to overcome external perturbations. We use a standard Genetic Algorithm (GA) for optimizing the synaptic weights of the ANN.