Space science missions (which includes all deep space missions) are funded for the acquisition of scientific data. Therefore, the return on investment can be measured by the quantity and quality of data retrieved. One realistic way to improve efficiency is to increase the amount of high quality science data that can be retrieved from missions. In this talk I will examine some of problems in obtaining high quality data from remote science missions and discuss work being carried out to improve the performance of remote probes and landers using on-board intelligent software. In particular, I will talk about planning systems and their role in autonomous system control. I will put the work in context by outlining the role it might play in a variety of planned missions and talk about some of the science drivers that are affecting the design for these missions.