{"title":"The interaction of maturational constraints and intrinsic motivations in active motor development","authors":"Adrien Baranes, Pierre-Yves Oudeyer","doi":"10.1109/DEVLRN.2011.6037315","DOIUrl":"https://doi.org/10.1109/DEVLRN.2011.6037315","url":null,"abstract":"This paper studies computational models of the coupling of intrinsic motivations and physiological maturational constraints, and argues that both mechanisms may have complex bidirectional interactions allowing the active control of the growth of complexity in motor development which directs an efficient learning and exploration process. First, we outline the Self-Adaptive Goal Generation - Robust Intelligent Adaptive Curiosity algorithm (SAGG-RIAC) that instantiates an intrinsically motivated goal exploration mechanism for motor learning of inverse models. Then, we introduce a functional model of maturational constraints inspired by the myelination process in humans, and show how it can be coupled with the SAGG-RIAC algorithm, forming a new system called McSAGG-RIAC2. We then present experiments to evaluate qualitative and, more importantly, quantitative properties of these systems when applied to a 12DOF quadruped controlled with 24 dimensions motor synergies.","PeriodicalId":256921,"journal":{"name":"2011 IEEE International Conference on Development and Learning (ICDL)","volume":"16 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2011-08-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116000194","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Towards incremental learning of task-dependent action sequences using probabilistic parsing","authors":"Kyuhwa Lee, Y. Demiris","doi":"10.1109/DEVLRN.2011.6037332","DOIUrl":"https://doi.org/10.1109/DEVLRN.2011.6037332","url":null,"abstract":"We study an incremental process of learning where a set of generic basic actions are used to learn higher-level task-dependent action sequences. A task-dependent action sequence is learned by associating the goal given by a human demonstrator with the task-independent, general-purpose actions in the action repertoire. This process of contextualization is done using probabilistic parsing. We propose stochastic context-free grammars as the representational framework due to its robustness to noise, structural flexibility, and easiness on defining task-independent actions. We demonstrate our implementation on a real-world scenario using a humanoid robot and report implementation issues we had.","PeriodicalId":256921,"journal":{"name":"2011 IEEE International Conference on Development and Learning (ICDL)","volume":"7 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2011-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114607955","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Giuseppe Cuccu, M. Luciw, J. Schmidhuber, F. Gomez
{"title":"Intrinsically motivated neuroevolution for vision-based reinforcement learning","authors":"Giuseppe Cuccu, M. Luciw, J. Schmidhuber, F. Gomez","doi":"10.1109/DEVLRN.2011.6037324","DOIUrl":"https://doi.org/10.1109/DEVLRN.2011.6037324","url":null,"abstract":"Neuroevolution, the artificial evolution of neural networks, has shown great promise on continuous reinforcement learning tasks that require memory. However, it is not yet directly applicable to realistic embedded agents using high-dimensional (e.g. raw video images) inputs, requiring very large networks. In this paper, neuroevolution is combined with an unsupervised sensory pre-processor or compressor that is trained on images generated from the environment by the population of evolving recurrent neural network controllers. The compressor not only reduces the input cardinality of the controllers, but also biases the search toward novel controllers by rewarding those controllers that discover images that it reconstructs poorly. The method is successfully demonstrated on a vision-based version of the well-known mountain car benchmark, where controllers receive only single high-dimensional visual images of the environment, from a third-person perspective, instead of the standard two-dimensional state vector which includes information about velocity.","PeriodicalId":256921,"journal":{"name":"2011 IEEE International Conference on Development and Learning (ICDL)","volume":"497 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2011-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133973713","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Realistic child robot “Affetto” for understanding the caregiver-child attachment relationship that guides the child development","authors":"H. Ishihara, Y. Yoshikawa, M. Asada","doi":"10.1109/DEVLRN.2011.6037346","DOIUrl":"https://doi.org/10.1109/DEVLRN.2011.6037346","url":null,"abstract":"Children are considered to develop various kinds of their social abilities in communication with their caregivers. Developmental researchers have revealed the quality of the caregiver-child attachment heavily affects the children's developmental passway and sometimes threatens their healthy development. For understanding developmental mechanism under the caregiver-child attachment, a number of theoretical models have been proposed and some child robots have been created to test these models or to find new facts about development. However, most of these robots have not been provided with a realistic childlike appearance or facial expressions, which seem important to induce caregiver's attachment. Since what kinds of treatment robots receive from the “caregivers” appears to depend on what kinds of impression the robots give to their caregivers, more realistic robot that is more close to a real child seems needed. In this paper, we introduce our project to build a new child robot, Affetto, that has realistic appearance of 1- to 2-year-old child, and discuss what kinds of issues on child development can be examined by using it.","PeriodicalId":256921,"journal":{"name":"2011 IEEE International Conference on Development and Learning (ICDL)","volume":"123 4 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2011-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129552179","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}