The Golem Group was founded by Dr. Luis A. Pineda in 1998 within the context of the project Diálogos Inteligentes Multimodales en Español (DIME, Intelligent Multimodal Dialogues in Spanish). This project was created with the objective of developing computational technology for the development of multimodal conversational agents with spoken Spanish and to work out a theory, a methodology and a programming environment that allowed the integration of different capacities in a software architecture by the reuse, the replacement and the addition of capacities and tasks in a simple manner.
Some of the result achieved were: the creation and transcription of corpus DIME, the development of a computational Spanish grammar, the creation of the “dialog model” concept, as well as the interpreter program for the creation of multimodal conversational systems and the integration of a flexible software platform for the construction of interactive systems with spoken Spanish.
In 2002 we developed the Golem Project and robot. First, we brought together the results of the DIME project; we integrated the technology of conversational and spoken recognition systems with different input and output modalities, like images and video, and the motor behavior of the robot for the construction of an intelligent and mobile agent able to interact within the context of multimodal structured conversations. In its initial version, the robot Golem was able to sustain a simple spoken conversation and to follow simple commands for movement.
In June 2007, Golem was presented at UNAM’s science museum, Universum, with a reliable dialogue manager and a robust Spanish speech recognizer. This new version of Golem was able to guide a poster session about the research projects of the Computer Science Department at IIMAS, and to interact visually with users by the recognition of gestures made with the arm. A video about this presentation and Golem’s functionality is available here.
In 2009, we incorporated computer vision capabilities to present in December the module “Guess the card: Golem in Universum”. From this date on, this application has been exhibited in this museum, where the general public, and especially children, can play a game about guessing a card chosen by the system.
Golem has caused great interest and has been widely covered by the Mexican media; TV, radio, and press and Internet and has been also widely demonstrated in several academic events in Mexico.
In 2010, we started the transformation of the Golem-II+ into a service robot, based on an innovative explicit cognitive architecture that, in conjunction with the dialogue models theory and the program interpreter constitute the theoretical core of our work. At that time, among other relevant results we integrated different capacities like the interpretation of pointing gestures during conversation, the reactive behavior during navigation and the orientation of the robot towards the source of sound to face the interlocutor in HR interaction, illustrating the capacity to interpret linguistic, visual and motor expressions in a coordinated way.
Late in 2010 we began developing the test scenarios of the RoboCup@Home competition to work up our theory, methodology and programming environment. From that date on the Golem Team has participated in national and international competitions; on April, 2012 we won the third place in the RoboCup German Open 2012, the 1st place in the Mexican Robotics Tournament and in June we obtained the 8th place in the RoboCup 2012.
