Human aware robot navigation
Kafle, Aayush (2022-06-30)
Kafle, Aayush
A. Kafle
30.06.2022
© 2022 Aayush Kafle. Ellei toisin mainita, uudelleenkäyttö on sallittu Creative Commons Attribution 4.0 International (CC-BY 4.0) -lisenssillä (https://creativecommons.org/licenses/by/4.0/). Uudelleenkäyttö on sallittua edellyttäen, että lähde mainitaan asianmukaisesti ja mahdolliset muutokset merkitään. Sellaisten osien käyttö tai jäljentäminen, jotka eivät ole tekijän tai tekijöiden omaisuutta, saattaa edellyttää lupaa suoraan asianomaisilta oikeudenhaltijoilta.
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:oulu-202206303203
https://urn.fi/URN:NBN:fi:oulu-202206303203
Tiivistelmä
Human aware robot navigation refers to the navigation of a robot in an environment shared with humans in such a way that the humans should feel comfortable, and natural with the presence of the robot. On top of that, the robot navigation should comply with the social norms of the environment. The robot can interact with humans in the environment, such as avoiding them, approaching them, or following them. In this thesis, we specifically focus on the approach behavior of the robot, keeping the other use cases still in mind. Studying and analyzing how humans move around other humans gives us the idea about the kind of navigation behaviors that we expect the robots to exhibit. Most of the previous research does not focus much on understanding such behavioral aspects while approaching people. On top of that, a straightforward mathematical modeling of complex human behaviors is very difficult. So, in this thesis, we proposed an Inverse Reinforcement Learning (IRL) framework based on Guided Cost Learning (GCL) to learn these behaviors from demonstration. After analyzing the CongreG8 dataset, we found that the incoming human tends to make an O-space (circle) with the rest of the group. Also, the approaching velocity slows down when the approaching human gets closer to the group. We utilized these findings in our framework that can learn the optimal reward and policy from the example demonstrations and imitate similar human motion.
Kokoelmat
- Avoin saatavuus [34546]