Start Date

9-5-2019 3:30 PM

End Date

9-5-2019 5:00 PM

Document Type

Event

Keywords

Humanoid, framework, model, AI, Nao

Description

High quality probabilistic models for a complex robot like Alde- baran’s Nao humanoid depend heavily on details from its envi- ronment, involving multiple parameters. Building such models re- quires significant effort with data gathering and data cleaning. We propose to create a public database of NAO sensor data that can be used by researchers and engineers training models for localization, mapping, and planning in controlled environments. Here we report on our release of a database with sensor data, parameterized by en- vironment configuration and nature. The database contains struc- tured folders with documentation, measurements, Nao AI feedback, and data management tools. The measurements can include transi- tion, sonar, and image data while the internal AI feedback contains results of deFlorida Atlantic Universitylt landmark detection.

DOI

https://doi.org/10.5038/JGEH3147

Share

COinS
 
May 9th, 3:30 PM May 9th, 5:00 PM

Towards Sensor and Motion Measurements Databases for Training Models for the NAO Robot

High quality probabilistic models for a complex robot like Alde- baran’s Nao humanoid depend heavily on details from its envi- ronment, involving multiple parameters. Building such models re- quires significant effort with data gathering and data cleaning. We propose to create a public database of NAO sensor data that can be used by researchers and engineers training models for localization, mapping, and planning in controlled environments. Here we report on our release of a database with sensor data, parameterized by en- vironment configuration and nature. The database contains struc- tured folders with documentation, measurements, Nao AI feedback, and data management tools. The measurements can include transi- tion, sonar, and image data while the internal AI feedback contains results of deFlorida Atlantic Universitylt landmark detection.