Takeaways from animal behaviors applied to robotics

From the study of animal behaviors, some major points were extracted that could be applied on the development of behaviors for robots:

  • Reflexive behaviors are “hardwired,” meaning that there can be a direct connection between sensory data and actions, reducing computation costs.
  • The internal state of the agent may have a rol on the release of behaviors, such as being hungry or being low on battery.
  • Behaviors can be innate (pre-programmed) or can be learned.
  • Behaviors can be sequenced to perform more complex behaviors.
  • We can use Boolean’s to simulate the IRM and control the release of behaviors.
  • Some behaviors may need only affordances for they release. This can help us reduce computational costs.
  • We can use schema theory 20211229200605 as a base for implementing behaviors using OOP.
  • Behavior are inherently distributed and concurrent.
  • The combination of different outputs of different behaviors is one of the mayor decisions. These can be canceled, combined, prioritized, etc.

Notes References

20211217210733 Behaviors as the building block of intelligence

sche

20211217202757 Robot behaviors

20211217203220 INDEX - Artificial Intelligence

References

(Murphy 2000)

Murphy, Robin. 2000. Introduction to AI Robotics. Intelligent Robotics and Autonomous Agents. Cambridge, Mass: MIT Press.