CREST Robotic Scientist

Test platform used to host the AI based Robotic Scientist software system
Test platform used to host the AI based Robotic
Scientist software system

The aim Robotic Scientist project is to allow a robotic vehicle such as a Mars rover to act as a surrogate for the science team back on Earth, by allowing it to detect scientific targets of interest and exploring these in greater detail without the need for detailed supervision from mission control. This autonomous, robotic scientist must be able to detect potential targets from sensors such as cameras using advanced image processing techniques. Once a target is detected it must choose an appropriate response which is compatible with the intent of the science team. For example it may simply be to take a high resolution image or move closer to the target in order to carry out more detailed analysis.
Having selected a desired action the system must then be able to decide whether or not it has sufficient resources or energy to carry out this unplanned procedure and ensure that it does not jeopardise the pre-planned science activities for the day. The robotic scientist will use intelligent planning and scheduling to carry out this task. The overall objective of the work is to maximise the quality and significance of the data returned to the science team for detailed expert analysis, as will be required of the ExoMars Rover.
Having proposed this concept to the Science and Technologies Facility Council under their CREST initiative, SciSys then led a team of industrialists and academics on a 12-month programme to develop and integrate the required technologies into a test bed demonstration. This successful demonstration was witnessed by ESA in a live trial and introduced a methodology for autonomous science assessment based on terrestrial field science practice. The components consisted of an autonomous, opportunistic science agent, a planning and scheduling system and an instrument placement agent.

View ‘The Autonomous Robotic Scientist’ video.

The overall objectives of this work were as follows:

  • Establish an initial scientific methodology for the automation of science assessment and planning based on terrestrial field practice
  • Prototype a system architecture which can support the concept of autonomous opportunistic science
  • Prototype elements of the methodology provided by the science team in order to establish the feasibility of this approach
  • Demonstrate the prototype system in a representative “Mars Yard” environment
  • Use the forthcoming ESA ExoMars mission as a target and source of operations and science requirements

Our partners in this work included Aberystwyth University, University of Leicester and the University of Strathclyde.

For further information, contact: Dr. Mark Woods at . See source link for more animations: