A new approach to onboard planning and scheduling for autonomous remote systems

In this paper we propose an onboard software framework for many types of remote systems with the capability of dynamic self-update without halting the software or shutting the hardware system, which are necessary for conventional onboard software updates. The key idea is to express an operational plan, which is usually a sequence of system-built-in commands, as a text script. This script contains not only the conventional plan but also “logic”-type commands so that the operational plan can be expressed in the form of a program fragment such as “IF A THEN DO B”. Once this logic-type command is received, a remote system can change its behavior without requiring the software to be rebooted. Using this software architecture, we propose a new search algorithm using an orthogonal experiment design in the onboard planning and scheduling process; orthogonal experiment designs have been used in many quality control processes. Heuristics is used in conventional applications to create options for what to do next in the planning process, whereas here the focus is placed on transforming the conventional use of the orthogonal experiment design to create sufficient number of optional plans for the system to test as small as possible. Finally, we report the result of a demonstration experiment using an autonomous underwater vehicle that is controlled by onboard software based on the proposed software architecture and the searching algorithm.