For a flying insect, estimating absolute ground speed is essential during behaviors such as tracking an odor plume, or keeping tracking of total distance travelled. Behavioral experiments indicate that insects are indeed capable of estimating absolute ground speed, however, the underlying mechanism remains unknown. The fundamental challenge is that insects only have access to measurements that provide relative information: vision provides a measure of optic flow which corresponds to the ratio of velocity and distance to nearby objects, whereas airspeed measurements provide a vector sum of ground velocity and wind velocity. To explore how an insect might decouple these relative measurements, we take a robotics-inspired-biology approach focused on the task of extracting absolute ground speed from noisy optic flow measurements. Mathematically, extracting absolute ground speed from optic flow requires non-zero accelerations, as well as estimating the time-derivative of optic flow. Through simulations and robotics experiments we find that there is a clear tradeoff in overcoming the challenge of differentiating noisy optic flow data. First, we show that errors in velocity estimates due to noise in the optic flow measurements can be reduced by spatially integrating optic flow across large receptive fields. Second, we find that in cluttered environments with multiple distinct objects at different distances, smaller receptive fields result in lower errors. Thus, for a given level of noise and environmental clutter, there exists an optimal size for receptive fields that balances this tradeoff. This observation may explain why flies have visual neurons with intermediate sized receptive fields, however, additional experiments are necessary to determine what the distribution of these neurons is.