Affordance discovery based on intrinsic motivation in robots

Affordance research is a new biology-inspired approach to promote the cognitive capabilities of robots. Affordances could capture the relationships between the robot and environment in terms of actions that the robot is able to perform. Intrinsic motivation is the fundamental factor that drives an o...

Full description

Saved in:
Bibliographic Details
Published in2016 IEEE International Conference on Robotics and Biomimetics (ROBIO) pp. 88 - 93
Main Authors Chang'an Yi, Huaqing Min, Jinhui Zhu, Pengshuai Yin
Format Conference Proceeding
LanguageEnglish
Published IEEE 01.12.2016
Subjects
Organisms
Planning
Probabilistic logic
Psychology
Robot kinematics
Search problems
Online AccessGet full text

Cover

More Information
Summary:Affordance research is a new biology-inspired approach to promote the cognitive capabilities of robots. Affordances could capture the relationships between the robot and environment in terms of actions that the robot is able to perform. Intrinsic motivation is the fundamental factor that drives an organism to learn even without external reward, and it is also the reason why an infant organism keeps interacting with the world and developing the knowledge. To search an object is the prerequisite to perceive later affordances such as graspable and pushable, and it is also the instinct action of an infant before 5 months. In this paper, we propose an intrinsic motivation based method to discover the searchable affordance for a robot equipped with limited perceptual and behavioral capabilities. The experiment is carried out in both simulation and real environments, where the NAO humanoid robot could construct the searchable relationships and make the target object be the center of its vision.
DOI:10.1109/ROBIO.2016.7866303