We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:

References & Citations

DBLP - CS Bibliography


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Robotics

Title: Perspective-corrected Spatial Referring Expression Generation for Human-Robot Interaction

Abstract: Intelligent robots designed to interact with humans in real scenarios need to be able to refer to entities actively by natural language. In spatial referring expression generation, the ambiguity is unavoidable due to the diversity of reference frames, which will lead to an understanding gap between humans and robots. To narrow this gap, in this paper, we propose a novel perspective-corrected spatial referring expression generation (PcSREG) approach for human-robot interaction by considering the selection of reference frames. The task of referring expression generation is simplified into the process of generating diverse spatial relation units. First, we pick out all landmarks in these spatial relation units according to the entropy of preference and allow its updating through a stack model. Then all possible referring expressions are generated according to different reference frame strategies. Finally, we evaluate every expression using a probabilistic referring expression resolution model and find the best expression that satisfies both of the appropriateness and effectiveness. We implement the proposed approach on a robot system and empirical experiments show that our approach can generate more effective spatial referring expressions for practical applications.
Comments: Accepted by IEEE Transactions on Systems, Man, and Cybernetics: Systems, 2022
Subjects: Robotics (cs.RO); Computation and Language (cs.CL); Human-Computer Interaction (cs.HC)
DOI: 10.1109/TSMC.2022.3161588
Cite as: arXiv:2104.01558 [cs.RO]
  (or arXiv:2104.01558v3 [cs.RO] for this version)

Submission history

From: Mingjiang Liu [view email]
[v1] Sun, 4 Apr 2021 08:00:02 GMT (10186kb,D)
[v2] Sun, 27 Jun 2021 09:02:27 GMT (10187kb,D)
[v3] Sun, 3 Apr 2022 06:46:16 GMT (2089kb)

Link back to: arXiv, form interface, contact.