Even if a socially interactive robot has perfect information about the location, pose, and movement of humans in the environment, it is unclear how this information should be used to enable the initiation, maintenance, and termination of social interactions. We review models that have been developed to describe social engagement based on spatial relationships and describe a system developed for use on a robotic receptionist. The system uses spatial information from a laser tracker and head pose information from a camera to classify people in a categorical model of engagement. The robot's behaviors are determined by the presence of people in these different levels. We evaluate the system using observational behavioral analysis of recorded interactions between the robot and humans. This analysis suggests improvements to the current system: namely, to put a stronger emphasis on movement in the estimation of social engagement and to vary the timing of interactive behaviors

The low-temperature adsorption of N(2) on Rh/SiO(2) samples of various particle-size distributions was followed by FTIR. The addition of O(2) pulses on Rh(0) surfaces saturated with chemisorbed N(2) allowed us to reassign stretching frequencies attributed originally to N(2)-Rh(0) to N(2)-Rh(delta+). The formation of the latter oxidized Rh species is assumed to be induced by an electron withdrawal from adsorbed oxygen species on Rh surface centers neighboring those onto which N(2) species are chemisorbed. The present work, thus, enables us to delimit ranges of frequencies for which the adsorption of N(2) can be considered to occur on either Rh(0) or Rh(delta+) centers for nu(N2) lower or higher than 2243 cm(-1), respectively. The N(2)-FTIR experiments performed on the studied catalysts also suggest a lattice plane selectivity for N(2) adsorption on metallic Rh planes of different natures which, to our knowledge, has not been reported yet for Rh.