The vast majority of today's AUVs operate with a pre-programmed mission plan specifying waypoints and vehicle parameters for the entire mission. Complex tasks that cannot be accurately specified in advance must be solved through intermittent communication with a human operator. This obviously limits the performance and applicability of such vehicles. A truly autonomous vehicle must be able to perceive its own condition and its environment, and respond appropriately to unexpected or dynamic situations. Updated situational awareness requires an extensive set of sensors and data analysis tools, but the most challenging part of decision autonomy is still to select advantageous actions based on the information available.
Conceptually, decision autonomy can be divided into two categories: The ability to handle internal malfunctions (sustainability) - and the ability to handle unpredictable external events (adaptivity). The latter is important for optimizing mission execution by adaptive, real-time mission planning based on e.g. observed bathymetry and sea current conditions. It also facilitates novel applications such as adaptive data collection and cooperation with other vehicles. Sustainability is vital to realise both long-endurance missions, as the probability of sub-system failures increases with mission duration, and missions in extreme environments, e.g. under ice, where consequences of failures may be unacceptable. In all cases, actions must be chosen so that the overall mission goals are achieved to highest degree possible. These actions may include modifying the mission plan, and algorithms for automated re-planning are thus required.
Section 3.1 provides an overview of the different architectures that are common to use when designing an autonomous control system. Section 3.2 introduces autonomous path planning, explaining the consequences of the control architecture on the complexity of the planning algorithms. Planning requires knowledge about the environment of the vehicle, and Section 3.3 discusses ways to represent this knowledge. For the vehicle to be truly autonomous, it is important that its sensors are also able to operate autonomously; this is discussed in Section 3.4. The ability to handle internal malfunctions is covered in Section 3.5. The autonomous vehicle may be required to cooperate with other vehicles, both autonomous and not. Some scenarios requiring such cooperation are given in Section 3.6. Finally, an example of an autonomous vehicle is given in Section 3.7, which very briefly describes the autonomy system of the HUGIN AUV.
Was this article helpful?