Computing design principles for robotic telescopes

Telescopes capable of making observing decisions independent of human supervision have become a reality in the 21st century. These new telescopes are likely to replace automated systems as the telescopes of choice. A fully robotic implementation offers not only reduced operating costs, but also significant gains in scientific output over automated or remotely operated systems. The design goals are to maximise the telescope operating time and minimise the cost of diagnosis and repair. However, the demands of a robotic telescope greatly exceed those of its remotely operated counterpart, and the design of the computing system is key to its operational performance. This paper outlines the challenges facing the designer of these computing systems, and describes some of the principles of design which may be applied. Issues considered include automatic control and efficiency, system awareness, robustness and reliability, access, security and safety, as well as ease-of-use and maintenance. These requirements cannot be considered simply within the context of the application software. Hence, this paper takes into account operating system, hardware and environmental issues. Consideration is also given to accommodating different levels of manual control within robotic telescopes, as well as methods of accessing and overriding the system in the event of failure.