The name AUDIOBOT is a concatenation of two radical taken from the words AUDIO and roBOT.

AUDIOBOT is envisaged as a robot with ears, natural hearing and listening behaviours, auditory capacity and audition sense to detect acoustics events, recognise activities and multi-sensory reasoning for interaction to provide high reliable information services for human collaboration and object cooperation.

AUDIOBOT acts as human listening capable of performing a variety of complex human tasks on command via voice/graphical user interfaces or programs via the platform.

The first research LISTENOR focus is on audition sense and auditory capacities, reception, natural behaviours and streaming of natural environmental sounds and noise.

The second research AUDIOBOT focus is on the capacity of processing of acoustics in order to identify events and recognise activities. This include two main areas: the audio-ontology development and the (provision of meaning) to the audio streaming in a protocol of audio graphical representation, categorisation and processing.

The third research SOUNDBOT focus is on sounds understanding to interact and collaborate for planned goals. This include the Bayesian processing and risk management and take decision tools.

Finally the four main research focus of AUDIOBOT is on the self-learning and external-training of the audition capacities of robots and robotics machineries. This include the Internet of Sound Architectures and Services of acoustics training.

AUDIOBOT benchmark the safe human-robot interaction and collaboration improvement by a better understanding of the external environment, for the major information and high quality of multi-sensory and contextual reasoning.

AUDIOBOT refers to an open standard audio-robot to provide reliable information supportive services for interaction and human collaboration and cooperation in Ambient Intelligent applications such as homes and/or remote monitoring services.

--------------- 

The consortium raised on before proposals and in the conference “State of the Art, Opportunities and Challenges of Acoustic Computing for Artificial and Ambient Intelligence Applications to provide world-wide innovative Information Society services” on Set09 in BCN

The community is multidisciplinary and involves the following technologies:

Artificial Audio Cognition systems; Audition sense and Auditory capacities for robotics; Audio-Sensing and Audio-Sensors; Audio (sounds-noises) computing (acquiring, processing, understanding, learning and reasoning); Audio-sensors networks; Acoustic Events Detection (AED); Audio representation and semantic classification; Internet of Sounds and services; Audio-sensors networks; Activity Recognition and goal collaboration; Audio-Services and Solutions and Social activities and other not mentioned;

Initial core team consortium:

and the following application for provide information (monitoring, reporting, informing, warming, alerting or alarming) assistive and supportive services for human:

Tele-monitoring and surveillance in Hearing remote temporal, moving and heterogeneous environments; Public and Personal Security and Privacy; Inclusion and Health; Mobility and Transit; Energy-efficiency; Logistics and Manufacturing and other domains and industrial sectors still not listed.

AUDIOBOT design, develop, evaluate and benchmark Artificial Audio Cognitive System able to learn and understand the meaning of sounds, reasoning activities and optimising reactions, in base of identify events and recognise contextual situations though acoustic processing in hearing and listening environments.

 ------------------

Main Work-packages

WP0 Project management and coordination (MUF)

WP1 Definition, concretion and detailed of works and objectives.

1) Involvement of end-users and experts for collaboration in the design, development and evaluation.

WP2 Audio-sensors and Audio acquisition, streaming and processing (IDMT)

1) Design and development of a interactive Mechatronic crown LISTENOR of two or three dual pair-microphones acting in two behaviours relaxed (omnidirectional) and inter-active (directional, tracking or focused on one point)

2) Design the audio-system to allow two main behaviours:

1. passive/relaxed for static hearing (omnidirectional sounds)

2. active/focused listening (addressing a specific or interesting audio points) such as a conversation or the main direction of a track or path when moving

3) Build a functional sounds and noises acquiring prototype, similar to plastic crown of two pair of double microphone situated in the opposite poles for fitting in the head, of an assistant robot.

The audio system, include the software to optimise the streaming, compilation, integration and multi-pre-processing of the audio-information acquired by the diverse microphones.

WP3 Audio meaning and understanding (representation and semantics) (IDMT)

Audio data analysis for situational awareness and detection of normal and abnormal activity. Data input from multiple microphones. Sounds and Event tracking.

Audio-Data must/can also be analysed for identifying and non-identifying activities and either saved as separate datasets or not saved. Audio-streaming data from previous time be “saved” depending on level of readiness, from minutes to days.

1) Design and Develop a graphical system to describe Acoustic Events as graph of sounds that has the information to be identify when the acoustic filtering process an audio stream;

2) Design, develop and implement a Semantic Library and Gallery of “Audio-Pictures” pieces of software that describe and filter a specific acoustic event.

3) Design and Evaluation of trust and reliability

WP4 Artificial Audio cognition system (AACS) for robotics

WP5 Multi-sensory interactions for collaboration with human

1) Developing syntax language for interacting with autonomy to build and develop a Bayesian framework for probabilistic control be able of take decision and extract solutions few prior structural knowledge in uncertain situations and complex dynamical environments and situation Assessment and conveying information within uncertainty

Enhancing User understanding of autonomous systems though the application of mental models

Designing architectures for complex autonomous systems (and integration of sub- systems)

Approaches to provide information to assist in design in Agent- based models

Knowledge Elicitation techniques (Semantic mapping) as tools to provide insight into building Agent behaviours and values

2) Development of autonomous robot teams to show collaborative behaviours with minimal user input, scaled for priority to allow dynamic goal- directed inject from user

WP6 Training and Learning Audition Sense (reasoning and interaction frameworks)

Learning Artificial Audio Cognition system to improve the quality of the audio perception of robots in order to recognise better the contextual situation and the understanding of tasks. We need a Internet of things and Services to training robots. The process initially designed could be collect all the abnormal sounds processed and feed in a internet semantic database to feedback the robots to upgrade software and improve the audition capacities (quality and quantity of activities recognized).

- Creating an artificial system, based on Internet services, capable of learning actions based on human audition sensory abilities to collect, process and classified AE to retrofitting the robot in upgrade packages

- Building a semi-driven system for deciding which actions are ore suitable in every situation of a collaborative tasks to training the robot. This will use collective behaviour arising from the interplay of individuals that recommends the “best action” for the current state of the tasks and combines hand-on trained systems and models models to guide the learning actions process

WP7 Benchmarking and evaluation for human collaboration in real-cases of task

1) Design Human Interaction for evaluation of human collaborative tasks

2) Drawn the situation and comparative assessment in different technological areas of the AUDIOBOT and robotics:

- Improvement of Positioning, Situation Aware and Trajectory Tracking capacities for autonomous Navigation in complex real environments

- Precision of the detection of events and recognition of activities and definition of the context

- Definition of contextual planned goals

3) Comparative evaluation of Reliability & Quality of service (mono – audio and video -, dual – audio plus video, and three senses – audio, video and gyroscope (speed perception) in specific collaborative tasks

4) Final and global assessment of human performance within Cognitive systems

WP8 AACS info-space and community for robotics and ICT solutions in application sectors

An Internet of Sounds and Services will be implemented to provide audio learning and training systems.

WP9 Dissemination and sustainable exploitation of AACS for industrial/sectoral applications

AUDIOBOT is opening and introducing an wide-world innovation and European “added-value” in the domain of sounds for robotics in particular in their capacities for listening, in the enhancement of robotics of recognition of events and understanding contextual situations. Additionally, this value refers to the speciality and multidisciplinary of the works and partners that cannot be found of the European level.

AUDIOBOT implies at least:

1) Leading-edge research capacity in Europe in audio cognitive systems engineering and robotics.

2) World-wide Innovation in service robots, and industrial production and manufacturing processes, in particular in the sector of telesurveillance, security and protection, Inclusion and Healthcare.

3) A new widespread comparative assessment of audition capacities for robot performance in different applications, collaborative tasks and technologies.

4) A new ICT of acoustics processing for AI/AmI market opportunities, and technologies and applications and services for increased productivity and efficiency in EU industries and administrations.

-- 
en Ferran Cabrer i Vilagut
Identity, diversity and complexity for sustainability
[CA] Cabrer.cat                                     Ferran @ Cabrer.cat
[EN] Cabrer.org                            Ferran.Cabrer @ MUFICATA.net 
_______________________________________________________________________
MUFICATA s.l. - Natural Information Environments
Scientific Prospectors, Technological Envisioners, Germinal Ideas, 
Conceptual Developers, Consortia Builders, Collaborative Netwokers, 
Innovative Performers, Responsible Managers and Sustainable Maintainers.
Skype/Google: Ferran.Cabrer        T:+34 93 423 8267   M:+34 6222 54458  
Jaume Fabra 12 08004 Barcelona (Catalonia)