embodied paradigm
Table of Contents
as opposed to the instinctive heuristic to consider an agent to have the world detected with sensor input, to:
consider agent to have the world detected with <experiment, sensor input> tuple.
In this way:
- Agent become part of it’s perception of the world.
- Agent learns about the world interactively
- Agent learns about how the world operates in addition to sensor readings. (which leads naturally to reasoning and decision making)
1. resource
Bibliography
[1]
O. L. Georgeon and A. Cordier, “Inverting the Interaction Cycle to Model Embodied Agents,” Procedia computer science, vol. 41, pp. 243–248, 2014, doi: 10.1016/j.procs.2014.11.109.
Backlinks
with a set of state \(S\) and a set of actions \(A\), the set \(S \times A\) refers to a set of 2-tuples \((s_t,a_t)\) where \(s_t \in S, a_t \in A\), enumerating all configurations, with the semantic of all possible interaction of agent with the environemnt
the notion of time spent in a subset of this set in here could be formulated as probability distribution of each interaction happening