Interest-Driven Exploration With Observational Learning for Developmental Robots

Intrinsic motivation learning for real robot applications

“…The results showed that novelty-based signals were more effective to drive the human learning. Still, there is no clear border between these two categories since high prediction errors indicate novel situations to learn from as shown recently in the novelty detection method (Rayyes et al, 2021). Similarly, (Barto et al, 2004; considered high prediction error as a novelty-based signal.…”

“…For instance, in (Forestier et al, 2017;Seepanomwan et al, 2017) the robot preformed random movements during the exploration which is inefficient and incompatible with humans'motion (von Hofsten, 2004). In contrast, (Tanneberg et al, 2018;Huang et al, 2019;Rayyes et al, 2021;2020a) have demonstrated high sample-efficiency and goal-directed motion. The only methods with notable high sample-efficiency are the methods which integrated intrinsic motivation with mental replay methods (Andrychowicz et al, 2017;Rayyes et al, 2020b).…”

“…In contrast to industrial robots, which accomplish predefined tasks, developmental robots must solve unpredictable tasks, learn new skills, adapt to new environments, and cope with unforeseen challenges. Intrinsic motivation methods tackle these challenges through driving the robot's learning and exploration autonomously by internally generated signals in an open-ended (i.e., unbounded) environment (Schmidhuber, 2010;Baranes and Oudeyer, 2013;Santucci et al, 2016;Baldassarre, 2019;Rayyes et al, 2020a;Rayyes, 2020;Rayyes et al, 2021). However, the high sample-complexity of these methods, i.e., the dense sampling required to approximate the learned function with a reasonable accuracy, restrict their real-world applications.…”

“…However, the high sample-complexity of these methods, i.e., the dense sampling required to approximate the learned function with a reasonable accuracy, restrict their real-world applications. Therefore, the majority of previous work has been demonstrated only in simulation as a proof of concept, and only a few were demonstrated in real robot experiments, e.g., (Tanneberg et al, 2018;Huang et al, 2019;Rayyes et al, 2020a;Rayyes et al, 2021).…”

Artificial cognition vs. artificial intelligence for next-generation autonomous robotic agents

“…The actual developmental level characterizes mental development retrospectively, while the Zone of Proximal Development characterizes mental development prospectively. For example, we may quote a representative list of studies in the two preceding decades that cover in isolation some of the specific issues of developmental robotics: Self-exploration and early imitation (Kuniyoshi et al, 2003), Modeling joint attention (Nagai, 2007), Scaffolding Robot Action Learning (Nagai and Rohlfing, 2009), Affordance-based perception (Min et al, 2016), Bootstrapping the semantics of tools (Schoeler and Wörgötter, 2016), Perception of Localized Features (Giagkos et al, 2017), Bootstrapping of Sensory-Motor Skills (Wieser and Cheng, 2018), Developing Reaching Ability like human infants (Luo et al, 2018), Sensorimotor Communication (Donnarumma et al, 2012(Donnarumma et al, , 2018Pezzulo et al, 2019), Integration of Sensing, Cognition, Learning, and Control (Li et al, 2019), Evaluation of Internal Models (Smith and Herrmann, 2019), Emergence of symbolic representations (Ugur et al, 2015;Taniguchi et al, 2019), Grounded affordances (Saponaro et al, 2020), Bodily Expression of Emotion (Tuyen et al, 2021), Morphological development (Naya-Varela et al, 2021), Skill Learning Strategy with Dynamic Motion Primitives (Li et al, 2021), Interest-driven exploration (Rayyes et al, 2022(Rayyes et al, , 2023.…”