SSC Pacific
 
Modular Robotic Architecture (ModBot) 

The Modular Robotic Architecture was developed by SSC San Diego as a generic platform control system offering developers a standard set of software and hardware tools that could be used to quickly design modular robotic prototypes with minimum start-up overhead (Smurlo & Laird, 1990). The concept facilitates customization of a testbed system by providing sensor, actuator, and processing modules that can be configured on demand as required by the particular needs of the application being addressed. The ability to later accept newer Modular Robotic Architecturemodules of increasing sophistication provides for evolutionary growth potential, ensuring maximum effective service life before the hardware becomes obsolete.

The ModBot is an example of a mobile robot implemented under this modular concept, employing several independent modules of varying intelligence and sophistication connected together in a generalized distributed network. The platform is made up of a detachable base with accompanying power source and various sensor, actuator, and processing modules. Each of these modules enables the robot to obtain and process different information about its surroundings.

The Collision Avoidance Sonar Module is active whenever the robot is in motion. It continuously looks for obstacles within a predefined distance and reports back to the High-Level Processing Module for appropriate action if an object is detected. The Near-Infrared Proximity Sensor Module is another means of determining if objects are in close proximity to the robot. This ring contains 11 Banner diffuse-mode optical proximity sensors facing the forward 180 degrees, each one having a range of approximately 3 feet. This module is used to complement data obtained by the Collision Avoidance Sonar Module. The High-Level Processing Module, housing a WinSystems AT286 computer mounted in a card cage, receives commands from the remote control station. This module uses its internal map representation, as well as information from other modules, to plan and execute a path to the desired destination.

During data transfers, the Control Station Module communicates with the ModBot via the Communications Module. An RS-232 umbilical cable was initially used during the early stages of development and later replaced by an OCI LAWN spread-spectrum RF link. Some exploratory work was also performed using a full-duplex near-infrared datalink made by STI. The modular nature of the robot allowed the Communications Module to be upgraded without any other reconfiguration necessary to the rest of the ModBot systems.

The flexibility and extendibility of the ModBot architecture have made it a valuable testbed for the pursuit of new ideas and applications involving robot mobility. One of the first was a significantly upgraded version of the robotic security concept carried over from ROBART II. ModBot TrackingThe Intrusion Detection Module is used to detect intruders in the vicinity of the robot and reports the bearing back to a remotely located Control Station Module. The Intrusion Detection Module consists of ultrasonic, passive-infrared, and microwave motion detectors which cover the full 360-degree surrounding area. A video motion detector in this module also receives information from the acoustic and video sensors on the Stereoscopic Pan-and-Tilt Module to determine if an intruder is present. Audio and composite video signals are transmitted back to the operator via two separate analog RF links.

ModBot's most recent role was a testbed for the Robotic Sensor-Motor Transformation research project, applying biological neural modeling to autonomous robotic control. For this application, another central processing module was added, housing an 80486 computer hosting two Intel i860 co-processors, and a 120MB hard disk. Sensory input consists of only data from the video camera and wheel encoders, all other sensory modules were removed. All processing as well as software development takes place on ModBot itself.

Bookmark and Share
Updated: 10/19/2011 9:40 PM EST   Published (2.0)