MultiON: Benchmarking Semantic mAP Memory Using Multi-Object Navigation

Abstract

Navigation tasks in photorealistic 3D environments are challenging because they require perception and effective planning under partial observability. Recent work shows that map-like memory is useful for long-horizon navigation tasks. However, a focused investigation of the impact of maps on navigation tasks of varying complexity has not yet been performed.

Cite

Text

Wani et al. "MultiON: Benchmarking Semantic mAP Memory Using Multi-Object Navigation." Neural Information Processing Systems, 2020.

Markdown

[Wani et al. "MultiON: Benchmarking Semantic mAP Memory Using Multi-Object Navigation." Neural Information Processing Systems, 2020.](https://mlanthology.org/neurips/2020/wani2020neurips-multion/)

BibTeX

@inproceedings{wani2020neurips-multion,
  title     = {{MultiON: Benchmarking Semantic mAP Memory Using Multi-Object Navigation}},
  author    = {Wani, Saim and Patel, Shivansh and Jain, Unnat and Chang, Angel and Savva, Manolis},
  booktitle = {Neural Information Processing Systems},
  year      = {2020},
  url       = {https://mlanthology.org/neurips/2020/wani2020neurips-multion/}
}