The problem of tracking and capturing a moving obstacle in two dimensions by a mobile robot equipped with a wide-beam sonar system is discussed. In a practical system the range and azimuth measurements contain random errors and are available over a limited region. This pursuer/prey problem is treated as a feedback control system for which the pursuer action is dependent on the observed measurements. Two measures of performance are considered: the capture probability and the mean capture time when capture occurs. The lower bound for the mean capture time is determined from game theory, which assumes complete information about the prey. Strategies employing either qualitative (prey is to the left or right) or quantitative (range and azimuth to prey) information are implemented and compared. It is found that qualitative information is sufficient for prey capture, although quantitative information allows more efficient prey capture.