This let Flakey drive along hallways with no dead reckoning or planning whatsoever. It seemed miraculous at the time; a situated automaton that knew things without needing any models. However, I thought of it as (sensor-driven) feedback control, versus (plan driven, eyes shut) feed-forward control. I then used Mike Georgeff's procedural reasoning system (PRS) to make Flakey not only drive but navigate an office building. In some respects this project succeeded: the robot's "domain knowledge" was nothing more than a static connection graph--no distances to drive, no widths of halls or doorways, no a priori obstacles--such information was acquired en route from sensory input.

The whole platform is 6 inches high; 12 inches wide, including wheels; and 14 inches long. The 68HC11 processor is in the middle, flanked by two side-pointing sonars. The rest of the sonars are in the front of the robot. The radio modem, with its antenna, is also on top. The housing contains a large battery and the motors.

The Second Annual Robotics Competition and Exhibition was held in July 1993 in conjunction with the National Conference on Artificial Intelligence. This article reports some of my experiences in helping to design and run the contest and some reflections, drawn from post mortem abstracts written by the competitors, on the relation of the contest to current research efforts in mobile robotics. The competition, which attracted teams from many of the top mobile robotics research laboratories in the United States (see side bar), was first proposed by Thomas Dean and held at the 1992 NCAI conference. Dean's concept was to further the research into the skills such robots need--sensing, interpretation, planning, and reacting--by bringing together interested parties in a cooperative and challenging environment. Ideas should be tested in the real world, not just the controlled conditions of the laboratory.

CARMEL's software design is hierarchical in Object recognition is done using a single camera and a one-pass algorithm to detect horizontally striped, bar-code-like tags on each of the 10 objects. A distance and a heading for each object are returned. Recalibrating the robot's position is done by triangulating from three objects with known locations. FLAKEY's basic software design is distributed: Articles On the perception side, modules add raw sonar and structured-light information to LPS, treating it as an occupancy grid. Other interpretive processes use this information to construct and maintain higher-order structures, parsing the data into surface segments, recognizing objects, and so on.

This design I thought I'd try something different. When I say built, I mean just servo couldn't turn fast enough to gather that: The students were given all the electronic enough data (the sonars can't be used while and mechanical components and, during the course, assembled and tested the circuit the servo is turning because they won't produce board (several smoky disasters here); built the reliable echoes). As a compromise, we chassis from bent aluminum sheet metal; used five sonars in the pattern shown in figure mounted motors, encoders, and wheels; and 3. The pivoting front sonar detects forward soldered a gazzillion cables to connect everything. I've had teaching a class.

The camera is mounted on a rotating table that allows it to turn 360 degrees independently of robot motion. Interestingly, the two teams processor (Z80) controls the robot's used vastly different approaches in the design wheel speed and direction. 's software design is hierarchical in The final scores for the robots, based solely structure. At the top level is a supervising on competition-day performance, constitute planning system that decides when to call only a rough evaluation of the merits of the subordinate modules for movement, vision, various systems. This article provides a technical or the recalibration of the robot's position.