Integratating Robotic Technologies with JavaBots

Mobile robotics research advances through developments in theory, and implementation in hardware and software. While theory is important, this article is primarily concerned with hardware and software technologies. It is our view that signiicant strides in robot performance can be made just by combining existing hardware and software tools. Thus the focus of this paper is answering the question: how can we more easily integrate robotic technologies? The typical research robot is a hodge-podge of sensor and actuator components from several vendors integrated in a common computing resource. Control system software accesses hardware through software APIs provided by the hardware developers, and \soft technologies" through their associated APIs. The focal point of integration is therefore the control system software. The problem with most control system implementations , from the standpoint of reusability and integration , is their committment to a robot platform and/or robot architecture. It is rather diicult for instance, to move a proven control system from one robot to another, because the APIs for sensing, steering and motor commands are often quite diierent in syntax and semantics on diierent robot hardware. Portabil-ity problems may extend to the software components of robotic systems as well: extracting the path planner from a robotic system, for example, is just as challenging as porting a control system. Therefore, to advance intelligent mobile robotics in the \Next Leap", we must emphasize integration of technologies from the multiple sub-areas of our eld. We should build systems by integrating, rather than re-inventing technology. Software frameworks for combining behavior, learning, planning, sensing and actu-ation should be inclusive rather than constrained to a particular robot architecture or hardware platform. We argue that the most eeective approach is through standardized interfaces (APIs) to robotic hardware and soft technologies (e.g. path planning toolkits). Jav-aBots is an example framework that provides this kind of integration in simulation and on robot hardware. A high-level common interface to sensors and actua-tors allows control systems to run on multiple simulated and real hardware platforms. Conversely, Jav-aBots supports the evaluation of competing control systems on the same hardware. In the rest of this article we brieey describe JavaBots, and provide examples of robot systems built using it. JavaBots JavaBots is a new system for developing and executing control systems on mobile robots and in simulation. Individual and multi-robot simulations are supported (including multiple robot types in the same simulation). The JavaBot system's modular design …