Embodied Navigation at the Art Gallery

Bigazzi, Roberto; Landi, Federico; Cascianelli, Silvia; Cornia, Marcella; Baraldi, Lorenzo; Cucchiara, Rita

doi:10.1007/978-3-031-06427-2_61

Cited by 4 publications

(2 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A variety of approaches leverage learning to overcome shortcomings in classical SLAM (Simultaneous Localization and Mapping) (c.f., [10]- [12]). For example, one can learn to generate classical SLAMstyle maps [13], topological maps [14], [15], multi-task deep memory representations [16], [17], and inferences over unseen regions [9], [18], [19]. In this work, we propose a learning-based approach to mapping semantic regions.…”

Section: Related Workmentioning

confidence: 99%

“…The online mapping experiment is performed using the Habitat simulator [46] with Mat-terport3D (MP3D) [48] environments. Among the available datasets for photo-realistic embodied navigation [51] like Gibson [52], Habitat-Matterport3D (HM3D) [53], only Matterport3D contains room annotations. We test our mapping approach on board of a state-of-the-art exploration method [9] to evaluate the agent's ability to correctly classify the observed regions.…”

Section: Online Mapping Experiments 1) Experimental Setupmentioning

confidence: 99%

See 1 more Smart Citation

Focus on Impact: Indoor Exploration With Intrinsic Motivation

Bigazzi¹,

Landi²,

Cascianelli³

et al. 2022

IEEE Robot. Autom. Lett.

View full text Add to dashboard Cite

Robots require a semantic understanding of their surroundings to operate in an efficient and explainable way in human environments. In the literature, there has been an extensive focus on object labeling and exhaustive scene graph generation; less effort has been focused on the task of purely identifying and mapping large semantic regions. The present work proposes a method for semantic region mapping via embodied navigation in indoor environments, generating a high-level representation of the knowledge of the agent. To enable region identification, the method uses a vision-tolanguage model to provide scene information for mapping. By projecting egocentric scene understanding into the global frame, the proposed method generates a semantic map as a distribution over possible region labels at each location. This mapping procedure is paired with a trained navigation policy to enable autonomous map generation. The proposed method significantly outperforms a variety of baselines, including an object-based system and a pretrained scene classifier, in experiments in a photorealistic simulator.

show abstract