CADDY – Cognitive Autonomous Diving budDY


Divers (SCUBA, scientific, and technical) operate in harsh and weakly monitored environments in which the slightest unexpected disturbances, technical malfunctions, or lack of attention of a diver can result in catastrophic consequences. These issues are usually dealt with by pairing up divers and adopting well defined rules for diving operations to reduce the chance of accidents. However, during more challenging dives these procedures may not be sufficient to ensure almost accident-free operations, for the divers must manoeuvre in complex 3D environments, carry cumbersome equipment, and focus attention on operational details.


The concept



autonomous surface robot




   FP7 / CP
   STARTING DATE – Jan 1, 2014
   ENDING DATE – Dec 31, 2016

COST € 742,824.00
FUNDING € 559,792.00


The core of the research and development effort will focus on setting up symbiotic links between a human diver and a set of companion autonomous robots. This motivates us to develop a multicomponent, highly cognitive underwater robotic system capable of learning, interpreting and adapting to the diver’s behaviour and physical state.

The core of the proposed envisioned concept consists of a diver, autonomous underwater robot and an autonomous surface robot. A diver will interact with the companion autonomous underwater robot who will manoeuvre underwater in the vicinity of the diver and exhibit cognitive behaviour with regard to the diver actions. The autonomous surface vehicle that communicates with the diver and the autonomous underwater robot is a communication relay link to the command centre, but at the same time it also plays the key role of a navigation aid to the underwater vehicles. It must adapt its motion so as to optimize the conditions for increased communications efficiency and navigational accuracy of the three components of the formation.

Assistance and monitoring in the CADDY sense refers to a threefold functionality similar to the one that a human buddy diver should have. Stated in simple terms, the system to be developed should play, among others, the simultaneous roles of: i) dive buddy “observer”, ii) dive buddy “slave”, and iii) dive buddy “guide”.
As a dive buddy “observer”, the system observes the diver at all times during the dive and interprets his(her) behaviour by assessing for example the body state, detecting the onset of nitrogen narcosis and signs of panic, and interpreting symbolic gestures communicated by the diver. The robotic buddy must manoeuvre safely around the diver in order to assume the best viewpoint for observation without interfering with the normal unfolding of the mission.
As a dive buddy “slave” the system affords the diver a “helping hand” to examine the environment; e.g. hovering over a spot indicated by a laser beam operated by the diver and taking photos of the location, following the diver and acquiring a series of overlapping photos for mosaic making, illuminating a site from different angles upon request from the diver, and carrying a payload with tools and equipment.
As a dive buddy “guide”, the system is in charge of actually guiding (upon request) the human diver from one spot to another, along a predefined search path, or steering the diver safely (in case of an emergency) to an appropriate point at the surface without violating basic diving rules and acting as an intelligent  communication router in situations where the diver loses line-of-sight to the surface vessel.

1. Develop a cooperative robotic system consisting of a surface and underwater vehicle that are capable of interacting with the diver.
2. Develop a reliable underwater sensing network that will enable recognition of diver pose and hand gestures.
3. Develop algorithms that will enable understanding of the diver behaviour, i.e. interpretation of both conscious (symbolic hand gestures) and unconscious (pose, physiological indicators) nonverbal communication cues.
4. Define and implement execution of cognitive guidance and control algorithms through cooperative formations and manoeuvres with a diver being a part of the formation.
5. Interpret complex sequences of diver gestures in order to achieve cognitive mission (re)planning.




University of Zagreb Faculty of Electrical Engineering and Computing, Croatia
National Research Council of Italy
Instituto Superior Técnico, Institute for Systems and Robotics, Portugal
Jacobs University, Germany
University of Vienna, Faculty of Life Sciences, Department of Anthropology, Austria
Newcastle University, School of Electrical & Electronic Engineering, UK
Diver Alert Network Europe, Malta