Humanoid infers Archimedes' principle: understanding physical relations and object affordances through cumulative learning experiences

Abstract

Emerging studies indicate that several species such as corvids, apes and children solve ‘The Crow and the Pitcher’ task (from Aesop's Fables) in diverse conditions. Hidden beneath this fascinating paradigm is a fundamental question: by cumulatively interacting with different objects, how can an agent abstract the underlying cause–effect relations to predict and creatively exploit potential affordances of novel objects in the context of sought goals? Re-enacting this Aesop's Fable task on a humanoid within an open-ended ‘learning–prediction–abstraction’ loop, we address this problem and (i) present a brain-guided neural framework that emulates rapid one-shot encoding of ongoing experiences into a long-term memory and (ii) propose four task-agnostic learning rules (elimination, growth, uncertainty and status quo) that correlate predictions from remembered past experiences with the unfolding present situation to gradually abstract the underlying causal relations. Driven by the proposed architecture, the ensuing robot behaviours illustrated causal learning and anticipation similar to natural agents. Results further demonstrate that by cumulatively interacting with few objects, the predictions of the robot in case of novel objects converge close to the physical law, i.e. the Archimedes principle: this being independent of both the objects explored during learning and the order of their cumulative exploration.

Why did I fail? A Causal-based Method to Find Explanations for Robot Failures

“…They find that excluding them impacts generalizability and sim-to-real transfer positively. In [11] a set of causal rules is defined to learn to distinguish between unimportant features in physical relations and object affordances. A humanoid iCub robot learns through cumulative experiences that dropping heavy objects into a jar of water will increase the water level, and other variables like color are irrelevant.…”

Transferring Adaptive Theory of Mind to Social Robots: Insights from Developmental Psychology to Robotics

“…We would like to urge future studies to focus on the modern cross-talk between developmental studies and robotics to answer these questions. In fact, on the one hand, developing architectures for robots inspired by developmental mechanisms resulted in more sophisticated robots with increasingly complex abilities and behavior [41]. On the other hand, robots have been useful in the modelling of human developmental processes within an embodied agent and the prediction of developmental phenomena which were successively validated by infants studies [2].…”

Humane Robots—from Robots with a Humanoid Body to Robots with an Anthropomorphic Mind

“…Although humans generally believe themselves to be focused on the here and now, the human brain is projected into the future: the brain is used to predicting the future (Berthoz 1997). We continuously imagine our actions and their potential effects, simulating our own movements through internal models that we learn and adapt to changes in our body (Bhat et al 2016). Thanks to the high similarity among conspecifics, the model of ourselves can represent a good approximation of the model we need to interpret and predict others.…”