Controlling limbs in robotics is a nonlinear, complicated calculation that animals can do without significant effort. In this work we present a dynamical neural model with bioinspired sensory receptive fields which is capable of encoding the forward kinematics of a robotic leg. The model is implemented using the SNS-Toolbox. Synaptic conductance values are tuned using the Functional Subnetwork Approach. No optimization or machine learning is required. To understand how network construction affects encoding accuracy, we systematically varied the sensory neuron receptor functions, the number of sensory neurons, and neuron time constants. We use the root-mean-squared error to check the accuracy of the designed model. Finally, we show that our model with multiple outputs is more efficient than multiple networks with one output each.