/Embodied-AI-Reactive-Hybrid-Architecture

ASEBA Programming Language for Thymio 2 Robot

Embodied-AI-Reactive-Hybrid-Architecture

"Intelligence always manifests itself in behaviour and that we must understand the behaviour." - Rolf Pfeifer

Author: Avinash Rai

Aseba Programming Language for Thymio 2 Robot

Notes:-

  • Purpose: To design, implement, experiment and optimize a self-sustaining agent (robot) tasked to clean garbage repeatedly from its environment with high efficiency.
  • Experimented in different unpredictible environments which will also include other agents serving to deter the primary agent from its goal (and survival).
  • An agent has only two primary sensors (front and ground) at its disposable to percieve its environment.
  • Agents have been designed with guidance from different Design Architectures defined in the Embodied Cognition literature.

Three main robots functionalities:-

MAIN BEHAVIOUR FILE DESIGN ARCHITECTURE
Wall following robot_1_wall_following.aesl Reactive - "Subsumption Architecture"
Garbage Collection robot_2_garbage_collection.aesl Hybrid - Motivated, Heirarchical and Reactive
Wandering robot_3_wanderer.aesl Reactive

Fig: Behaviour Action Selection Dynamics for Garbage Collection action_selection_dynamics

Two environments:-

  1. simple_environment.playground
  2. complex_environment.playground