Human pair walking behavior: evaluation of cooperation strategies*

Human walkers are notoriously poor at keeping a direction without external cues: Experimental work by Souman {\it et al.} with blindfolded subjects told to walk in a straight line revealed intriguing circular and spiraling trajectories, which can be approximated by a stochastic process. In this work, motivated by pair walking experiments by Miglierini {\it et al.}, we introduce an analysis of various strategies employed by a pair of blindfolded walkers, who are communicating via auditory cues, to maximize their efficiency at walking straight. To this end, we characterize pairs of strategies such as free walking, side-by-side walking and unconditional following from data generated by robot pair walking experiments (using computer vision techniques) and numerical simulations. We extract the mean exit distances of walker pairs from a corridor with finite width to construct phase portraits of the walking performance. We find intriguing cooperative effects leading to non-trivial enhancements of the efficiency at walking straight.