A Biologically Inspired Approach for Robot Depth Estimation

Figure 6

Samples of our architecture’s performance when the depth reference system is defined as Δ_r = {[0, 60[, [60, 100[, [100, 120[, [120, 250[, [250, ∞[} and Q = {very_close (vc), close (c), nearby (n), far (f), very_far (vf)}, while the orientation relationships are coded according to the Freksa and Zimmermann’s approach [48] (i.e., LAB₀ = {front_left (fl), straight_front (sf), front_right (fr), left (l), none (n), right (r), back_left (bl), straight_back (sb), back_right (br)} and INT₀ = {]90, 180[, [90, 90], ]0, 90[, [180, 180], _, [0, 0], ]180, 270[, [270, 270], ]270, 360[}).