This paper discusses concepts developed under the SEA led UK MoD Battlespace Access Unmanned Underwater Vehicle (BAUUV) programme to provide higher levels of military AUV autonomy. BAUUV aims to identify and assess the technology readiness relating to a range of future (2010-2015) UK military missions and to perform focussed technology development activities to address key technology gaps. One key gap relates to provision of suitable levels of autonomy to allow a UUV to perform long duration military missions. Contemporary autonomous underwater vehicles generally execute a prescripted mission plan with simple branching. However, future military missions will require higher levels of autonomy such that the vehicle can operate with a minimum of supervision and adapt to changing military goals, onboard health and situational awareness. In order to reduce onerous and often impracticable human supervision and communications requirements, future military vehicles will need to perform some level of autonomous mission replanning and decision making in order to adapt to changes in AUV situational awareness, changes in knowledge of vehicle status and energy availability and changes in military goals. Autonomous mission replannning algorithms that aim to provide this functionality have been developed and evaluated. The resulting mission replanning software utilises a hierarchical iterative approach with initial rough planning based on goal selection and sequencing activities being followed by detailed task planning and plan tuning. Replans can be instigated by user defined changes in goal characteristics or priorities or by internal triggers such as an unexpected change in energy usage or task progress. In addition to core mission replanning algorithms, software relating to specific task planning/replanning modules is being considered. For example, an autonomous transit task planner has been developed. This is capable of autonomously defining and costing a transit based on encyclopaedic knowledge of subsurface currents, detection and physical risks. The AUV “personality” is defined by the relative vehicle energy, risk and time priorities which drive the selection of a particular transit plan. Other task planners being considered include those relating to communications, survey, reconnaissance, REA and logistics goals. Within a typical three-layer UUV hierarchical control architecture, the onboard mission and task replanning elements would form part of the top level deliberative elements and would typically interface to a sequencing layer via an updatable mission script. The sequencing layer would interface to task achieving behaviours and low level autopilot modes, potentially via a collision and obstacle avoidance module. In addition to onboard elements, this paper discusses associated concepts relating to intuitive user interfaces and planning aids. Goal based planning/replanning technology enables the user to specify a mission based on a series of military goals, constraints and priorities rather than having to define a detailed mission script. This should increase the speed for the definition, validation and modification of future missions and reduce the skill requirement for a future military UUV user. An example goal based user interface prototype is presented. Finally, after describing current study results and status, the paper will touch upon ongoing trial activities to advance the technology from TRL 4/5 to TRL 6.
[1]
Wolfram Burgard,et al.
Experiences with an Interactive Museum Tour-Guide Robot
,
1999,
Artif. Intell..
[2]
Joan Batlle,et al.
Recent trends in control architectures for autonomous underwater vehicles
,
1999,
Int. J. Syst. Sci..
[3]
Keith Golden,et al.
Autonomous rovers for Mars exploration
,
1999,
1999 IEEE Aerospace Conference. Proceedings (Cat. No.99TH8403).
[4]
James Albus.
4D/RCS: A Reference Model Architecture for Unmanned Vehicle Systems
,
2002
.
[5]
Nicola Muscettola,et al.
Automated Planning and Scheduling for Goal-Based Autonomous Spacecraft
,
1998,
IEEE Intell. Syst..
[6]
Junku Yuh,et al.
Real-Time Control Architecture for Sauvim 1
,
2003
.
[7]
Steve Ankuo Chien,et al.
Autonomous Science on the EO-1 Mission
,
2003
.
[8]
G. Rabideau,et al.
CASPER: using local search for planning for embedded systems
,
2001
.
[9]
Alex Fukunaga,et al.
Iterative Repair Planning for Spacecraft Operations Using the Aspen System
,
2000
.
[10]
Rodney A. Brooks,et al.
A Robust Layered Control Syste For A Mobile Robot
,
2022
.
[11]
Rebecca Castano,et al.
The Techsat-21 autonomous sciencecraft constellation
,
2001
.
[12]
Tara Estlin,et al.
The CLARAty architecture for robotic autonomy
,
2001,
2001 IEEE Aerospace Conference Proceedings (Cat. No.01TH8542).
[13]
Ramon Abel Castano,et al.
Learning and planning for Mars Rover science
,
2003
.