Name
Abstract | ||
|---|---|---|
Robotic middlewares increasingly allow the seam- less integration of multiple heterogeneous robots into one distributed system. Unfortunately, very simple devices like tagged everyday objects and smart objects are left orphan in this otherwise pervasive trend. We claim that the inclusion of simple everyday objects as part of distributed robot systems would have many advantages, and propose a design pattern to allow this inclusion. We make this pattern concrete by describing an implementation of it using a specific multi-ro bot middleware, called PEIS-Ecology Middleware. We also show an illustrative experiment which integrates everyday objects in a smart home equipped with mobile robots as well as more advanced distributed sensor nodes. I. I NTRODUCTION Distributed multi-robot systems are becoming increasingly common in the robotic field (1), (2), (3). These systems often rely on some middleware to provide a common abstraction of the robotic devices in the system, as well as communi- cation mechanisms that allow these devices to seamlessly exchange data and commands. Many of these middlewares allow heterogeneous robots and devices to be part of one and the same system, provided that they have the required computation and communication resources. Robots in a distributed robot system can access properties and send commands among each-other via this middleware. However, if they need to access the properties of simpler ob- jects and devices that lack the required resources, they have to resort to some other means. For instance, a robot equipped with an RFID tag reader can access the information stored in the RFID tag attached to a parcel; and a robot equipped with a ZigBee interface can turn on a lamp controlled by a simple ZigBee based module. The parcel and the lamp, however, remain external to the distributed robotic system. The starting point of this paper is the realization that it would be extremely useful to make everyday objects and simple devices part of a distributed robot system, by making them accessible throughout the system via the same middleware (4). Consider for instance a transport robot T perceiving a smaller robot S, which is part of the same distributed system. By using techniques such as anchoring (5), T can realize that its perception of S and the digital representation of S in the distributed system refer to the same physical object. Thus T can acquire needed physical properties of S by querying its digital representation through the middleware — e.g., its grasping points, in order to be grasped. Suppose now that C is a coffee cup with an RFID tag that stores, among other things, its grasping points. If C could be accessed by T through the middleware in the same way as S was, then T would have a simple uniform way of addressing the grasping task. Today's robotic middlewares are usually limited to be run on relatively powerful processors (6), (7) and cannot incorpo- rate very small devices or everyday objects. Some previous works have shown robotic middlewares that with help of a light-weight/tiny version can also incorporate simple dev ices, like microcontrollers or wireless sensor network motes (7), (8), (9), (10). Although simple, these devices are assumed to be able to run software capable of communicating and interfacing with the general middleware. In this paper, we go one step further, and we propose a technique to include in a distributed system even objects and devices which are unable to run this software. These may be, for instance, everyday objects attached to RFID tags, or augmented with simple sensors and small communication modules (e.g., based on infrared communication or ZigBee) (11). Our technique is based on the notion of proxy: a process hosted by a component of the distributed robot system, which acts as a representative of the simple object inside the middleware. The proxy maintains an image of the external object which is made accessible to the middleware, and it uses a dedicated communication channel to synchronize this information with the actual object (e.g., an RFID reader, or a ZigBee radio module). In the next section, we elaborate the above idea into a general design pattern. We then show an implementation of this pattern using a specific multi-robot framework, called PEIS-Ecology. Finally we describe an experiement, which incorporates everyday objects in the P EIS-Ecology using the proposed technique, as a proof of concept, and conclude. |
Year | DOI | Venue |
|---|---|---|
2008 | 10.1109/IROS.2008.4651056 | Nice |
Keywords | Field | DocType |
control engineering computing,distributed processing,middleware,multi-robot systems,PEIS-ecology middleware,digital representation,distributed robot system,multiple heterogeneous robots,multirobot middleware,robot ecology | Middleware,Robotic systems,Computer science,Home automation,Smart objects,Robot,Mobile robot,Design pattern,Distributed computing,Information and Computer Science | Conference |
ISBN | Citations | PageRank |
978-1-4244-2057-5 | 3 | 0.54 |
References | Authors | |
8 | 3 | |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Rashid, J. | 1 | 3 | 0.54 |
| Mathias Broxvall | 2 | 301 | 25.54 |
| Alessandro Saffiotti | 3 | 2755 | 284.17 |