Creating autonomous robots that can actively explore the environment, acquire knowledge and learn skills continuously is ultimate achievement envisioned in cognitive developmental robotics. Importantly, if aim to create develop through interactions with their learning processes should be based on physical social world manner of human development. Based this context, paper, we focus two concepts models predictive coding. Recently, have attracted renewed attention as a topic considerable...

One of the biggest challenges in robotics is interacting under uncertainty. Unlike robots, humans learn, adapt, and perceive their body as a unity when with world. Here, we investigate suitability <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">active inference</i> , computational model proposed for brain governed by free-energy principle, robotic perception action nonsimulated environment. We designed deployed algorithm on humanoid iCub...

Active inference is a mathematical framework which originated in computational neuroscience as theory of how the brain implements action, perception and learning. Recently, it has been shown to be promising approach problems state-estimation control under uncertainty, well foundation for construction goal-driven behaviours robotics artificial agents general. Here, we review state-of-the-art implementations active state-estimation, control, planning learning; describing current achievements...

Recent advances in neuroscience have characterised brain function using mathematical formalisms and first principles that may be usefully applied elsewhere. In this paper, we explain how active inference—a well-known description of sentient behaviour from neuroscience—can exploited robotics. short, inference leverages the processes thought to underwrite human build effective autonomous systems. These systems show state-of-the-art performance several robotics settings; highlight these...

In this letter, we propose a complete probabilistic tactile-based framework to enable robots autonomously explore unknown workspaces and recognize objects based on their physical properties. Our consists of three components: 1) an active pretouch strategy efficiently workspaces; 2) touch learning method learn about properties (surface texture, stiffness, thermal conductivity) with the least number training samples; 3) algorithm for object discrimination, which selects most informative...

The predictive functions that permit humans to infer their body state by sensorimotor integration are critical perform safe interaction in complex environments. These adaptive and robust non-linear actuators noisy sensory information. This paper introduces a computational perceptual model based on processing enables any multisensory robot learn, update its configuration when using arbitrary sensors with Gaussian additive noise. proposed method integrates different sources of information...

The field of motor control has long focused on the achievement external goals through action (e.g., reaching and grasping objects). However, recent studies in conditions multisensory conflict, such as when a subject experiences rubber hand illusion or embodies an avatar virtual reality, reveal presence unconscious movements that are not goal-directed, but rather aim at resolving conflicts; for example, by aligning position person’s arm with embodied avatar. This second, conflict-resolution...

Human decisions are increasingly supported by decision support systems (DSS). Humans required to remain "on the loop," monitoring and approving/rejecting machine recommendations. However, use of DSS can lead overreliance on machines, reducing human oversight. This paper proposes "reflection machines" (RM) increase meaningful control. An RM provides a medical expert not with suggestions for decision, but questions that stimulate reflection about decisions. It refer data points or suggest...

We address self-perception in robots as the key for world understanding and causality interpretation. present a mechanism that enables humanoid robot to understand certain sensory changes caused by naive actions during interaction with objects. Visual, proprioceptive tactile cues are combined via artificial attention probabilistic reasoning permit discern between inbody outbody sources scene. With support exploiting intermodal contingencies, can infer simple concepts such discovering...

We present a pixel-based deep active inference algorithm (PixelAI) inspired by human body perception and action. Our combines the free energy principle from neuroscience, rooted in variational inference, with convolutional decoders to scale directly deal raw visual input provide online adaptive inference. approach is validated studying action simulated real Nao robot. Results show that our allows robot perform 1) dynamical estimation of its arm using only monocular camera images 2)...

This paper formulates and proposes a discrete solution for the problem of finding lost target under uncertainty in minimum time (Minimum Time Search). Given searching region where some information about is known but uncertain (i.e. location dynamics), agent with constrained dynamics, we provide two decision making algorithms that optimizes actions to find time. The faced as optimization: sensor are discrete, probabilistic model described over graph, each vertex contains target's probability...

The development of breakthrough technologies helps the deployment robotic systems in industry. implementation and integration such will improve productivity, flexibility competitiveness, diverse industrial settings specially for small medium enterprises. In this paper we present a framework that integrates three novel technologies, namely safe robot arms with multi-modal auto-calibrated sensing skin, control to generate dynamic behaviors fusing multiple sensor signals, an intuitive fast...

Self/other distinction and self-recognition are important skills for interacting with the world, as it allows humans to differentiate own actions from others be self-aware. However, only a selected group of animals, mainly high order mammals such humans, has passed mirror test, behavioural experiment proposed assess abilities. In this paper, we describe process that is built on top body perception unconscious mechanisms. We present an algorithm enables robot perform non-appearance...

This paper proposes a Bayesian approach for minimizing the time of finding an object uncertain location and dynamics using several moving sensing agents with constrained dynamics. The exploits twice theory: on one hand, it uses formulation objective functions that compare paths other optimization algorithm to solve problem. By combining both elements, our handles successfully this complex problem, as illustrated by results over different scenarios presented statistically analyzed in paper....

The minimum time search in uncertain domains is a searching task, which appears real world problems such as natural disasters and sea rescue operations, where target has to be found, soon possible, by set of sensor-equipped searchers. automation this the detect critical, can achieved new probabilistic techniques that directly minimize Expected Time (ET) dynamic using observation probability models actual observations collected sensors on board selected technique, described algorithmic form...

Humans can experience fake body parts as theirs just by simple visuo-tactile synchronous stimulation. This body-illusion is accompanied a spatial drift in the perception of real limb towards limb, suggesting an update estimation resulting from work compares drifting patterns human participants, rubber hand illusion experiment, with end-effector displacement multisensory robotic arm enabled predictive processing perception. Results show similar both and robot experiments, they also suggest...

In this paper, we introduce the main components comprising action-perception loop of an overarching framework implementing artificial attention, designed to fulfil requirements social interaction (i.e., reciprocity, and awareness), with strong inspiration on current theories in functional neuroscience. We demonstrate potential our framework, by showing how it exhibits coherent behaviour without any inbuilt prior expectations regarding experimental scenario. Current research cognitive systems...

Abstract The perception of our body in space is flexible and manipulable. predictive brain hypothesis explains this malleability as a consequence the interplay between incoming sensory information expectations. However, given interaction action, we might also expect that actions would arise due to prediction errors, especially conflicting situations. Here describe computational model, based on free-energy principle, forecasts involuntary movements sensorimotor conflicts. We experimentally...

In this paper we discuss the enactive self from a computational point of view and study suitability current methods to instantiate it onto robots. As an assumption, consider any cognitive agent as autonomous system that constructs its identity by continuous interaction with environment. We start examining algorithms learn body-schema enable tool-extension, finalize studying their viability for generalizing model. This points out promising techniques bodily self-modelling exploration, well...