Design for navigation

Directed Sonar Sensing for Mobile Robot Navigation

“…Many failed to realize that the physics of sonar dictates that all reflections produced by smooth planar targets are in fact specular reflections. (Thus our insistence that multiple reflections should not be called "specularities" [39], [123].) Although hardware details are not given, it seems clear to us that Kuc and Brown use transmitted pulses of short duration, and thus weak returns are not produced by their hardware.…”

“…To remove false multiple reflections, some researchers have artificially added acoustic targets to the environment. For example, Steer has developed acoustic retro-reflectors, that a user could place along the smooth surfaces of a room to provide a large number of corner targets and thus make false multiple reflections unlikely [123]. One drawback of such a technique is that modifying the environment may be undesirable.…”

Bee-havior in a mobile robot: the construction of a self-organized cognitive map and its use in robot navigation within a complex, natural environment

“…As with all sensors, sonar range data is subject to uncertainties which must be properly understood in order for the data to be used in a meaningful fashion. Due to specular reflection, a true range cannot be reliably obtained from these devices when measuring an acoustically smooth surface at an angle of incidence more than 1/2 of the beamwidth [13]. Two of the most common ways to combat this sonar uncertainty include (i) engineering the environment so as to eliminate the problem (e.g., using acoustic retroreflectors as in [13]) or (ii) building a model of the uncertainty expected.…”

“…Due to specular reflection, a true range cannot be reliably obtained from these devices when measuring an acoustically smooth surface at an angle of incidence more than 1/2 of the beamwidth [13]. Two of the most common ways to combat this sonar uncertainty include (i) engineering the environment so as to eliminate the problem (e.g., using acoustic retroreflectors as in [13]) or (ii) building a model of the uncertainty expected. The latter approach requires the robot to identify landmarks by matching their range signatures to stored templates [14].…”