The cognitive robotics laboratory carries out interdisciplinary research centered in the study of cognition and cognitive processes. The lab was funded in 2010 and it is part of the Center for Science Research (Centro de Investigación en Ciencias, CinC) in the Universidad Autónoma del Estado de Morelos (UAEM).
Cognitive Robotics focuses its attention on the design of artificial agents capable of performing cognitive tasks autonomously. A central issue in this consists in studying process by which agents learn through interaction with their environment. Cognitive Robotics aims to implement models of cognitive processes coming from Cognitive Sciences. The guidelines in this research area are a direct response to the shortcomings of Classical Artificial Intelligence, where high-level tasks and behaviors were studied. Our work is based on the concept of low-level sensorimotor schemes coded by Internal Models, thus falling as a matter of course within the tenets of Embodied Cognition, particularly with the idea that cognition must be understood as occurring in agents that have a body with which they interact in a specific environment. It is through this interaction that learning emerges laying the ground for cognitive processes. Our research includes theoretical work laying the foundations of Embodied Cognitive Robotics, as well as work with artificial and with natural agents.
The main theoretical concept behind the work in the lab is Internal Models.
Internal Models have been shown to be a good strategy to acquire the regularities of the environment as they are constructed by learned multimodal associations.
The notion of Internal Models implies that any model for motor adaptation knows something about its motor apparatus and the environment. In the study of Internal Models, forward and inverse models have been proposed, and studied in many other disciplines, such as neuroscience, biology and philosophy. Each of these disciplines presents different ontological commitments to the notion of Internal Models. In Embodied Cognitive Robotics, care has been taken to use only their core capabilities, namely, their ability to fuse multimodal information and provide predictions.
An inverse model, or controller, generates a motor command to the motor plant and a forward model predicts the output of the system, that is, the sensory consequences based on the motor command. A forward model incorporates knowledge about the sensory changes that will be produced by self- generated actions of an agent. In other words, the forward model predicts the sensory situation S*t+1 given a motor command Mt applied to an initial sensory situation St. Thus, forward models play a central role in cognitive agents as they provide predictions of the sensory consequences of motor commands.
The interest on internal models have taken the lab to dive into the ideas proposed by the predictive processing framework. This framework presents a new and groundbreaking theory on how perception, cognition and learning develop in natural agents.
The main research areas in the lab are:
Research aiming at implementations of models for cognitive processes on artificial agents. Among these processes we work with perception, sense of agency and knowledge acquisition. Cognitive processing depends on the sensorimotor cap- abilities an agent possesses, based on its body, situated in a specific environment. Our main research interest is centered on the understanding of how agents acquire, modify, and improve their sensorimotor schemes during interaction with the environment to accurately predict the sensory consequences of self-generated actions.
In order to model cognitive processes we do research on natural agents. This includes the study of established models as well as the development of new ones. Currently we are particularly focused on:
sense of agency,
influence of task and context for action and attention,
temporality in internal models.
The main aim of this line is to find out how relevant the theory of predictive processing can be for cognitive robotics. This theory has had an impact in all the topics that we study.
In this line of research the main aim is to use the principles of biological evolution to implement neuro-controllers for artificial agents. In particular we are interested on the emergence of communication and altruism among artificial agents.