ORCA: A physics based, robotics simulator able to distribute ...

More documents

Recommendations

Info

Robot models Using the aforementioned threee components, a developer can instantiate any type of f robotic platform. However, ORCA comes with a number of pre-defined structures for some popular robots (Fig. 3), includingng the Pioneer 3-AT platform, the iRobot3ADm, the Kuka LWR and a generic hand and arm model with 27-DoF. national projects. As a result of f these projects, the package was extended to include severall add-ons. The first class of add-ons regards human-robot communication technologies t (Fig. 4). These include components for natural n language interaction and virtual emotions, as well as HRI interfaces for human-robot interaction. Figure 3. (left) The simulated iRobotB21R. (right) The simulated Pioneer 3-AT platform. In addition, we are working towards including other platforms, such as the NAO and Hoap3 humanoids, the Kuka mobile platform and the PR2 robot. These will be made available upon ORCA’s next major release, andd will be optimized to reflect the physical properties of the actual hardware robotic platforms. Interaction Behaviors One of the main benefits of ORCA is that it defines explicitly how objects and physical entities will interact with each other. This is supported through NDE’s ability to define explicit materials on each object. Properties of the interaction include the use of gravity, collision mode, convex collision tolerance, net force limit, linear damping, translation and rotational impulse velocity. Moreover, each behavior can be set to run based on different collision primitives, be it mesh, or primitive bounding boxes, and can assign different mass, density values to interacting objects. Finally, torque and collision estimation can be performed locally for each behavior, through the implementation of appropriate callback functions. One important benefit from this feature is that it gives complete control on the physical representation of objects in the world. One can use it to describe how different materials will interact together, and as a a result define explicitly cross-object interactions, instead of describing a set of simple physics properties, as it i is the case with other simulators. One important result from this feature, is that one can turn off and on the physics processing of an object on demand, by freeing its assigned behaviors at any time, thus improving the speed of simulationn by eliminating redundant entities. In the video that accompanies the publication we demonstrate an example in which we make use of this property, to segment the physics processing of a simulation environment into different workstations. C. Add-ons ORCA has already been used in a large number of national and international projects, ncluding the EC- running FP7 project FIRST-MM, ass well as several funded FP6 projects GNOSYS and INDIGO, the currently Figure 4. Add-ons developed for project purposes. (left) Speech synthesis module (right) HRI interface for robot control through a touchscreen. More specifically, this family of add-ons for synthesizing speech and phrases, simply by typingg in the desired text, (ii) a includes threee modules; (i)) an application virtual emotion module, m whichh provides an interface for drawing 2D template faces from standard emotions and (iii) a GUI interface that allows humans to t control the robots by typing on o a touch screen. The second class of add-ons that have been developedd pertains to locomotion modules. More specifically, ORCA includes specific controllers that enable mobile robots to perform localization, obstacle avoidance, and a mapping. These modules, despite running in their own instance, interact with ORCA’s server communication system, in order to read sensor measurements and dispatch control commands to the robots being simulated (Fig. 5). Figure 5. Mapping and localization add-ons developed for project purposes. Finally, ORCAA is currently being expanded in the FP7 project FIRST-MM, which aims at developing novel components for mobile m manipulation. In the course of the project, ORCA is being extended to includee visual object tracking modules, as well as additional components thatt allow for instructing robots to perform guidance, delivery and transportationn tasks. IV . RESOURCE SHARING One of the strong benefits of ORCA iss its ability to share resources amongst different simulation instances. To cope with this requirement, ORCA has been designed to
handle all environment processing through its server object. Consequently it can tackle the large computational costs that are inherent in complex robotic structuress and large environments, by sharing and distributing resources across simulation instances. This functionalityy is embodied through threee different properties in n the simulator’s environment: (i) robot sharing, (ii) environment sharing, (iii) object sharing. Thesee are outlined below. A. Robot sharing One of the main characteristics of a simulated robot is that it utilizes resources that are modular, and can be described by an I/O convention. Consequently, in most cases, rendering a simulated robot is a computationally intensive task that outputs only a small set s of parameters. Consider for example the case where a “high degrees of freedom” manipulator has to be processed within a large environment. At any given time, the simulatorr has the duty to process its kinematic chain, integratee the appropriate equations, and update the position of the joints of the hand to simulate it. Even though this process contains a large number of computations, it is only used to output a small amount of information, , such as the end- with such a robot, it is usually important to have onlyy this point location of the hand. Consequently, when interacting piece of information available. Due to its design, ORCA can facilitate processing of such roboticc structures locally, and distribute outputs from the simulation on demand, only when required (Fig. 6). simulation environments or being transmitted by the server. B. Environment sharing Another important feature of the ORCA simulator is its ability to share the processing requirements of large environments. Such feature is becoming increasingly important across research projects, and finds applications in a large number of fields including outdoor robotics, search and rescue simulations, swarm robots etc. In these typess of environments, usually the robot is located within a small manifold space, which must process. Upon initialization off an experiment, the serverr loads a virtual world file whichh contains the definitions of all objects that will be processed by simulation instances. These are createdd as physical entities and entered into the world without processing any of their properties (Fig. 7). Figure 7. (right) The environment-sharing view of thee server showing how different robots are being processed using different subsets of an environment. (left) Rendered with redd is robot 1, green robot 2, while transparently are rendered the surroundings that are not n processed by any robot. As soon as a new simulation instance is i loaded into the new world, it communicates its location to the server. The server responds to the request of thee instance by transmitting the environment and object properties thatt are in the vicinityy of the roboticc client (Fig. 8). Figure 6. The environment-sharing view of thee server showingg how different robots are being processed using ORCA’s virtual robots. The simulation instances transmit all required infromation to a server, which for the given example, reduces the robot’s existence to a spatial x, y, z vector. To facilitate this feature, the latest ORCA release includes an extensive network protocol that contains messages pertaining to information sharing amongst robots. Thesee messages are integrated within appropriate environment parsers, thatt reside on the server andd are responsible for filtering and transmitting to simulation instances only the appropriate environment information. In addition, the latest release uses virtual robots, simple graphic structures with no physical properties, that can be used to visualize information coming from other Figure 8. Local copy of o the simulator that receives onlyy a partial view of the environment.
Page 1 and 2: ORCA: A physics based, robotics sim
Page 3: produced and consumed respectively

ORCA: A physics based, robotics simulator able to distribute ...

You also want an ePaper? Increase the reach of your titles

Delete template?

Save as template?