FLOWERS Research team
Can a machine learn like a child? Can it learn new skills and new knowledge in an unknown and changing environment? How can an embodied agent, e.g. a robot, discover its body and its relationships with the physical and social environment? How can its cognitive capacities continuously develop without the intervention of an engineer? What can it learn through natural social interactions with humans?
These are the questions that are being investigated in the FLOWERS research team at Inria Bordeaux Sud-Ouest and Ensta ParisTech. Rather than trying to imitate the intelligence of adult humans like in the field of Artificial Intelligence, we believe that trying to reconstruct the processes of development of the child's mind will allow for more adaptive, more robust and more versatile machines. This fundamental approach to the challenge of autonomous learning is called developmental robotics, or epigenetic robotics, and integrates concepts and theories from artificial intelligence, machine learning, neuroscience and developmental psychology. As many theories in neuroscience and developmental psychology are not formalized, this implies a crucial computational modeling activity, which in return provides means to assess the internal coherence of theories and sketch new hypothesis about the development of the human child's sensorimotor and cognitive abilities. Such computational modelling is also used as a foundational conceptual basis to build flexible lifelong autonomous machine learning systems.
Our team focuses in particular on the study of developmental constraints that allow for efficient open-ended learning of novel sensorimotor and interaction skills in embodied systems. In particular, we study constraints that guide exploration in large sensorimotor spaces:
Mechanisms of intrinsically motivated exploration and active learning, including artificial curiosity, allowing to learn diverse skills in the absence of any external rewards, and in particular to self-organize developmental trajectories (also called automated curriculum learning) and collect efficiently learning data;
Mechanisms of adequately constrained optimization and statistical inference for sample efficient sensorimotor skill acquisition (e.g. for optimizing motor policies in real robots through few interactions with the real world);
Mechanisms for social learning, e.g. learning by imitation or demonstration, which implies both issues related to machine learning and human-robot interaction;
Constraints related to embodiment, in particular through the concept of morphological computation, as well as the structure of motor primitives/muscle synergies that can leverage the properties of morphology and physics for simplifying motor control and perception;
Maturational constraints which, coupled with the other constraints, can allow the progressive release of novel sensorimotor degrees of freedom to be explored;
We also study how these constraints on exploration can allow a machine to bootstrap multimodal perceptual abstractions associated to motor skills, in particular in the context of modelling language acquisition as a developmental process grounded in action.
Among the developmental principles that characterize human infants and can be used in developmental machines, FLOWERS focuses on the following three principles:
Exploration is progressive. The space of skills that can be learnt in real world sensorimotor spaces is so large and complicated that not everything can be learnt at the same time. Simple skills are learnt first, and only when they are mastered, new skills of progressively increasing difficulty become the behavioural focus;
Internal representations are (partially) not innate but learnt and adaptive. For example, the body map, the distinction self/non-self and the concept of “object” are discovered through experience with initially uninterpreted sensors and actuators, guided by experience, the overall pre-determined connection structure of the brain, as well as a small set of simple innate values or preferences.
Exploration can be self-guided and/or socially guided. On the one hand, internal and intrinsic motivation systems regulate and organize spontaneous exploration; on the other hand, exploration can be guided through social learning and interaction with caretakers.
The work of FLOWERS is organized around the following axis:
Curiosity-driven exploration and sensorimotor learning: intrinsic motivation are mechanisms that have been identified by developmental psychologists to explain important forms of spontaneous exploration and curiosity. In FLOWERS, we try to develop computational intrinsic motivation systems, and test them on embodied machines, allowing to regulate the growth of complexity in exploratory behaviours. These mechanisms are studied as active learning mechanisms, allowing to learn efficiently in large inhomogeneous sensorimotor spaces and environments;
Cumulative learning of sensorimotor skills: FLOWERS develops machine learning algorithms that can allow embodied machines to acquire cumulatively sensorimotor skills. In particular, we develop optimization and reinforcement learning systems which allow robots to discover and learn dictionaries of motor primitives, and then combine them to form higher-level sensorimotor skills.
Natural and intuitive social learning: FLOWERS develops interaction frameworks and learning mechanisms allowing non-engineer humans to teach a robot naturally. This involves two sub-themes: 1) techniques allowing for natural and intuitive human-robot interaction, including simple ergonomic interfaces for establishing joint attention; 2) learning mechanisms that allow the robot to use the guidance hints provided by the human to teach new skills;
Discovering and abstracting the structure of sets of uninterpreted sensors and motors: FLOWERS studies mechanisms that allow a robot to infer structural information out of sets of sensorimotor channels whose semantics is unknown, for example the topology of the body and the sensorimotor contingencies (propriocetive, visual and acoustic). This process is meant to be open-ended, progressing in continuous operation from initially simple representations to abstract concepts and categories similar to those used by humans.
Body design and role of the body in sensorimotor and social development: We study how the physical properties of the body (geometry, materials, distribution of mass, growth, ...) can impact the acquisition of sensorimotor and interaction skills. This requires to consider the body as an experimental variable, and for this we develop special methodologies for designing and evaluating rapidly new morphologies, especially using rapid prototyping techniques like 3D printing.
Intelligent Tutoring Systems:FLOWERS develops methods for online personalization of teaching sequences for educational software and MOOCs. This work builds on top of online optimization methods and motivational research previously developed.