ORO, a knowledge management platform for cognitive architectures in robotics

Lemaignan, Séverin; Ros, Raquel; Mösenlechner, Lorenz; Alami, Rachid; Beetz, Michael

doi:10.1109/iros.2010.5649547

Cited by 127 publications

(97 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…For instance, the Oro framework creates a sophisticated ontology and uses it to justify observed inconsistencies on a robot [26]. The KnowRob architecture for service robots uses knowledge bases created from different sources to perform limited analysis of the reasons for unexpected observations [27].…”

Section: Related Workmentioning

confidence: 99%

Towards an Explanation Generation System for Robots: Analysis and Recommendations

2016

View full text Add to dashboard Cite

A fundamental challenge in robotics is to reason with incomplete domain knowledge to explain unexpected observations and partial descriptions extracted from sensor observations. Existing explanation generation systems draw on ideas that can be mapped to a multidimensional space of system characteristics, defined by distinctions, such as how they represent knowledge and if and how they reason with heuristic guidance. Instances in this multidimensional space corresponding to existing systems do not support all of the desired explanation generation capabilities for robots. We seek to address this limitation by thoroughly understanding the range of explanation generation capabilities and the interplay between the distinctions that characterize them. Towards this objective, this paper first specifies three fundamental distinctions that can be used to characterize many existing explanation generation systems. We explore and understand the effects of these distinctions by comparing the capabilities of two systems that differ substantially along these axes, using execution scenarios involving a robot waiter assisting in seating people and delivering orders in a restaurant. The second part of the paper uses this study to argue that the desired explanation generation capabilities corresponding to these three distinctions can mostly be achieved by exploiting the complementary strengths of the two systems that were explored. This is followed by a discussion of the capabilities related to other major distinctions to provide detailed recommendations for developing an explanation generation system for robots.

show abstract

Section: Related Workmentioning

confidence: 99%

Towards an Explanation Generation System for Robots: Analysis and Recommendations

2016

View full text Add to dashboard Cite

show abstract

“…The usual approach to this problem consists in developing some dedicated feature extractors such as shape or color descriptors [10], which are fed to a dimensionality reduction technique such as bag-of-features [77]; their output is matched with some pre-existing, handcrafted symbolic knowledge such as ontologies, extracted from large databases or from the Internet [86], which can be used for cognitive planning [47]. In these paradigms, robot actions are used more to match robot sensory signal to high-level abstractions instead of being used to explore in the sensory space to automatically discover these abstractions, reducing the need for prior models of the environment.…”

Section: Introductionmentioning

confidence: 99%

Deep unsupervised network for multimodal perception, representation and classification

Droniou

Ivaldi

Sigaud

2015

Robotics and Autonomous Systems

View full text Add to dashboard Cite

In this paper, we tackle the problem of multimodal learning for autonomous robots. Autonomous robots interacting with humans in an evolving environment need the ability to acquire knowledge from their multiple perceptual channels in an unsupervised way. Most of the approaches in the literature exploit engineered methods to process each perceptual modality. In contrast, robots should be able to acquire their own features from the raw sensors, leveraging the information elicited by interaction with their environment: learning from their sensorimotor experience would result in a more efficient strategy in a life-long perspective. To this end, we propose an architecture based on deep networks, which is used by the humanoid robot iCub to learn a task from multiple perceptual modalities (proprioception, vision, audition). By structuring high-dimensional, multimodal information into a set of distinct sub-manifolds in a fully unsupervised way, it performs a substantial dimensionality reduction by providing both a symbolic representation of data and a fine discrimination between two similar stimuli. Moreover, the proposed network is able to exploit multimodal correlations to improve the representation of each modality alone.

show abstract

“…In the work of Lemaignan [38] information inference using an online server is presented through a practical demonstration. Lim [40] subsequently establishes a multi-level representation of robot and environment knowledge, where Bayesian inference and heuristics are used to have a robot complete an under-informed task.…”

Section: Introductionmentioning

confidence: 99%

Cloud based centralized task control for human domain multi-robot operations

Janssen

Molengraft

Bruyninckx

et al. 2015

Intel Serv Robotics

View full text Add to dashboard Cite

With an increasing number of assistive robots operating in human domains, research efforts are being made to design control systems that optimize the efficiency of multi-robot operations. As part of the EU funded RoboEarth project, this paper discusses the design of such a system, where a variety of existing components are selected and combined into one cohesive control architecture. TheNote to Practitioners The work described in this article is part of the EU funded RoboEarth project, in which knowledge repositories and intelligent services are developed to operate service robots world wide. The centralized task controller described here can be regarded as one of these intelligent services, that realizes a time-optimal allocation of multiple robots. This work shows significant similarities with the (not publicly available) ubiquitous task controller designed by Ha, where our work uses a more expressive action language. The general use of our system from a human end-user point of view is to request for a list of tasks, for which an abstract plan and according parameterizations are selected by the system. The selected plan is subsequently parsed into a Golog based planning interpreter, that binds available robots and all the other parameters (such as objects and locations). The parametrized actions involved in the plan are mapped onto computational algorithms and robots in a time-optimal manner, and subsequently executed by interfacing with them through a service(http)-based interface. What should be noted is that the executing robots themselves serve merely as sensor-actuator platforms, leaving all the computations to be performed on the Cloud. Our opinion is that this deployment advances the opportunity for calculation optimization and the instant reuse of updated task knowledge, e.g. object locations, action plans, and robot capabilities on a global level. architecture's main design principle stems from Radestock's 'separation of concerns', which dictates the separation of software architectures into four disjunct components; coordination, configuration, communication and computation. For the system's coordinating component a Golog based planning layer is integrated with a custom made execution module. Here, the planning layer selects and parametrizes abstract action plans, where the execution layer subsequently grounds and executes the involved actions. Plans and plan related context are represented in the OWL-DL logics representation, which allows engineers to model plans and their context using first-order logic principles and accompanying design tools. The communication component is established through the RoboEarth Cloud Engine, enabling global system accessibility, secure data transmissions and the deployment of heavy computations in a Cloud based computing environment. We desire these computations, such as kinematics, motion planning and perception, to all run on the Cloud Engine, allowing robots to remain lightweight, the instant sharing of data between robots and other algorithms and most impo...

show abstract

ORO, a knowledge management platform for cognitive architectures in robotics

Cited by 127 publications

References 9 publications

Towards an Explanation Generation System for Robots: Analysis and Recommendations

Towards an Explanation Generation System for Robots: Analysis and Recommendations

Deep unsupervised network for multimodal perception, representation and classification

Cloud based centralized task control for human domain multi-robot operations

Contact Info

Product

Resources

About