Energy-aware fault-mitigation architecture for underwater vehicles

• Published in: Autonomous robots Vol. 41; no. 5; pp. 1083 - 1105
• Main Authors: De Carolis, Valerio, Maurelli, Francesco, Brown, Keith E., Lane, David M.
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.06.2017; Springer Nature B.V
• Subjects: Actuators; Architecture; Artificial Intelligence; Autonomous underwater vehicles; Autonomy; Computer Imaging; Control; Energy management; Engineering; Fault detection; Fault tolerance; Inspection; Level (quantity); Mechatronics; Parameters; Pattern Recognition and Graphics; Performance degradation; Representations; Robotics; Robotics and Automation; Subsystems; Vehicles; Vision; Fault tolerance; Energy awareness; Autonomous underwater vehicle; Knowledge representation; Automatic fault-mitigation
• ISSN: 0929-5593; 1573-7527
• DOI: 10.1007/s10514-016-9585-x

Energy awareness and fault tolerance are important aspects for extending the autonomy levels of today’s autonomous vehicles. With the aim of preparing the way for persistent autonomous operations of underwater vehicles this work focusses its efforts on investigating the effects of actuator failures, on an autonomous underwater vehicle (AUV) capable of long-term inspection missions. This paper introduces an energy-aware architecture that by observing the use of the on-board resources is capable of detecting faults and monitors the performance of the thruster subsystem in modern AUVs. The effect is an increased autonomy level in presence of unexpected events like performance degradations or sudden failures. Moreover an important contribution of this work is to process the great volume of information, collected at the lower sensor levels, into operational parameters that can be treated by higher level modules. These parameters form part of an abstract representation of concepts and capabilities that are available at a given time during the mission’s execution. Once this representation has been updated it is made available to the planning and execution components that can adapt the mission’s behaviour using the most recent knowledge about the vehicle’s state. To validate the proposed approach we evaluate our system on a real platform, Nessie VIII AUV, in both in real sea conditions and in a controlled test tank.