Dataset for spatial concept formation
/CC4F_turtlebot2/: Creation-Core Building (4th floor) in Ritsumeikan University, using TurtleBot 2/map/: Files on an environmental map (including formats supported by ROS)- Training data: position and words
/SIGVerseV2/: Home enviornment (One room) on SIGVerse version 2/map/: Files on an environmental map/stream_data/: stream data of observation and latent variables for MCL- Training data: position and words
/albertB/: albert-B-laser-vision-dataset from the Robotics Data Set Repository (Radish)/map/: Files on an environmental map (including formats supported by ROS)- Training data: position, words and image fetures
/img_feature/: Image features obtained using a pretrained CNN/places205/: Places205-AlexNet/places365/: Places365-ResNet
NEW
/SIGVerseV3/: Home enviornments on SIGVerse version 3/similar/3LDK_small/: Small dataset in 10 similar home enviornments
Speech signal data is not included in the dataset.
These data sets are used in the following papers:
- Akira Taniguchi, Tadahiro Taniguchi, and Tetsunari Inamura, "Spatial Concept Acquisition for a Mobile Robot that Integrates Self-Localization and Unsupervised Word Discovery from Spoken Sentences", IEEE Transactions on Cognitive and Developmental Systems, Vol. 8, No. 4, pp.285-297, 2016.
- Akira Taniguchi, Tadahiro Taniguchi, and Tetsunari Inamura, "Unsupervised Spatial Lexical Acquisition by Updating a Language Model with Place Clues", Robotics and Autonomous Systems, Vol. 99, pp.166-180, 2018.
- Akira Taniguchi, Yoshinobu Hagiwara, Tadahiro Taniguchi, and Tetsunari Inamura, "Online Spatial Concept and Lexical Acquisition with Simultaneous Localization and Mapping", IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS2017), pp.811-818, Sep, 2017. in Vancouver, Canada.