Figure 6

Expected sensation. Plots of the expected sensation of the edge and corner feature for all trajectories. The axes are associated with the sensor measurements; 0 means that the corresponding feature is not sensed and 1 the feature is fully sensed. A point in the plot summarise a whole trajectory by the mean and variance of the probability of sensing a corner or edge. The radius of the circles are proportional to the variance. The dotted blue rectangle represents the decision boundary for classifying a trajectory as being either risk-prone or risk-averse. A point which lies inside the rectangle is risk-prone. Left: Human trajectories demonstrate a wide variety of behaviours ranging from those remaining close to features to those preferring more risk. Right: Red points show greedy and blue points the GMM model. Bottom: Green circles are associated with the hybrid method, whilst orange are those of the coastal navigation method. The hybrid method is a skewed version of the GMM which tends towards risky behaviour and exhibits the same kind of behaviour as the coastal algorithm.