Space Robotics at Aberystwyth University

Space Robotics Research Focus

• Robot enabled space and planetary science.
• Autonomous science sample/data acquisition.
• Advanced computer vision methods for space and planetary robots.
• Robot and camera calibration.

Past and Current Mission Involvement, and Future Mission Studies

• Beagle 2 - Aberystwyth University (AU) involvement included: ARM simulation, calibration and mission commanding.
• ESA/NASA ExoMars 2018 - AU involvement includes: PanCam Instrument, Rover Inspection Mirror (RIM), PanCam Calibration Target (PCT), ROCC Mars Terrain Simulator (MTS).
• Aurora Core Programme - Involved in EXploration Robotics (X-Rob) study, and Mars Surface Sample Transfer and Manipulation (MSSTM) study.
• BeagleNET - A future Mars mission based upon Beagle 2 heritage. AU responsible for the simulation studies.
• Lunar Beagle - A future Lunar mission based upon Beagle 2 heritage. AU involvement includes 3D vision, and control of the Instrument Arm (IA).
• MoonLITE - Involved in the descent camera and 3D vision processing.
• ESA Cosmic Vision Marco Polo - Responsible for Near Earth Asteroid (NEA) sample acquisition simulation.

Space Robotics Projects Gallery

• The Beagle 2 Development Model (DM) undergoing trials at the Beagle 2 Lander Operations Control Centre (LOCC), National Space Centre, Leicester. The arm is moving the Beagle 2 mole towards the Gas Analysis Package (GAP) inlet port. A screen shot of the AU arm commanding simulation software can be seen in the background. i-SAIRAS 2003 paper here (PDF).

  

• Members of the LOCC team performing Wide Angle Mirror (WAM) tests. The Beagle 2 Stereo Camera System (SCS) was mounted on the Beagle 2 Position Adjustable Workstation (PAW) which was at the end of the robot arm. This meant that when one of the contact instruments was being used to collect data, the SCS could be pointing sky-wards. When deployed, the WAM popped up into the field of view of one of the cameras. Thus by imaging the WAM a panoramic image could be captured. The bright spot in the image is a passive marker that is reflecting light from a flash-unit.

  

• AU Beagle 2 surface shadowing simulator (movie here). The movie shows a speeded up sol (Martian day) from dawn till dusk. During the Beagle 2 mission it was important to know where the robot arm would cast shadows. For example, engineers did not want to position the arm so that shadows would be cast on the solar panels.

  

• AU BeagleNET simulation (movie here). The movie shows the deployment of the solar panels, mast/arm, and the egress of the four-wheeled rover. The rover deploys a seismometer from its body (to measure Marsquakes), and then moves on to obtain surface and sub-surface samples. Upon return to the lander, the mast/arm transfers the samples from the rover to the lander Gas Analysis Package (GAP).

  

• AU Lunar Beagle arm commanding study. The arm is the Beagle 2 Development Model (DM) arm and the project was in collaboration with EADS Astrium Ltd. (movie here). The movie shows the Beagle 2 arm being un-stowed and executing straight line motion to deploy selected instruments relative to a predefined target. Lunar and Planetary Science XXXIX 2008 paper here (PDF).

  

• AU 'deformable' robot arm (movie here). The arm was designed and built as part of our Model-Based Kinematics (MODKIN) project. The arm linkages can be 'twisted' and 'bent' so as to emulate severe arm damage. The MODKIN project focused upon developing autonomous fault diagnostic and damage remediation techniques. A new Fuzzy Trigonometry has been developed and used to model the arm's kinematics. This allows Fuzzy Denavit-Hartenberg parameters to be used to represent the kinematics of a damaged arm. The project was in collaboration with Aberdeen University. Link to Int. J. Approximate Reasoning 2009 paper here.

  

• The Autonomous Robotic Scientist project was funded by the UK STFC Collaborative Research in Exploration Systems and Technology (CREST) initiative. The rover contained advanced planning and scheduling-based tactical replanning software, called time line validation and control (TVCR); a science assessment and response agent (SARA), and an arm agent and perception interface (AAPI). The combined software was able to process camera captured images, identify rocks with a bedding structure, for example, and plan and schedule the activities needed to traverse towards the science target, capture and process stereo images for the 3D data required by the arm kinematics, and place the robotic arm against the identified science target. The project was in collaboration with SciSys Ltd., and the universities of Leicester and Strathclyde. Link to Journal of Field Robotics 2009 paper here.

  

• AU Multi-spectral PanCam emulator undergoing field trials with the EADS Astrium Ltd. 'Bridget' rover. Presentation here (PDF). Field trials have been conducted at a quarry near Stevenage, and in July 2010 at our local Clarach Bay beach. This event was reported in a UK Space Agency news item.

  

• AU Knowledge-based Science Target Identification System (KSTIS) prototype user interface. KSTIS can process PanCam images to generate a science value rank order for each rock in an image. KSTIS is an 'expert system' that applies domain expert knowledge (from a planetary geologist) to assess each candidate rock within an image to identify the 'best' science target. KSTIS employs three fuzzy rule-bases: structure, texture and composition; one responsible for each of the three examined rock attributes. These rule-bases utilise Mamdani's fuzzy inference method, a number of membership functions, and a collection of rules. The combined output is then de-fuzzified using Centre of Gravity (COG) defuzzification. This returns a 'crisp' number which represents a rock's science value. IROS 2009 paper here (PDF).

  

• AU Shape from Shading (SFS) algorithm example. Left is the original 2D Mars Express HRSC H1022 image, right is the AU SFS 3D result that has been rendered under opposite lighting conditions. A new optimisation based shape from shading algorithm has been developed at AU which is able to make use of sophisticated camera and reflectance models. Accurate results have been obtained when the algorithm has been applied to both synthetic rendered surfaces, and real images captured by the ESA Mars Express orbiter and HRSC instrument. The generated surfaces provide improved fine surface detail over that found using shape from stereo techniques. Link to ISPRS Journal of Photogrammetry and Remote Sensing 2012 paper here.

  

• AU aerobot undergoing acceptance trials at the ESA ESTEC Planetary Testbed Facility. The project focused upon autonomous image-based localisation for a future Mars aerobot mission. Project partners included SciSys Ltd., the University of Leicester and Joanneum Research, Austria. ISPRS 2006 paper here (PDF).

  

• AU autonomous cooperating aerobots (movie here), The speeded-up movie shows our aerobots demonstrating formation flying. Each aerobot is autonomous and has a behaviour-based controller. An external Vicon motion tracking system uses a wireless network to let each aerobot know the position of itself and its partners in Cartesian space. When an unexpected visitor entered the room, an air-draft perturbed the aerobots, but their behavioural controllers were able to correct for this disturbance. The project focused upon autonomous cooperant control methods for a terresrial application, and we are keen to explore this technology for a future Mars multi-aerobot mission. Project partner SciSys Ltd. Link to Robotica 2009 paper here.

  

• AU has studied the use of fuzzy-rough feature selection (FRFS) to aid efficient and accurate Mars terrain image classification. Using FRFS allows the induction of low-dimensionality feature sets from sample descriptions of feature vectors of a much higer dimensionality. Comparative studies have demonstrated that FRFS helps to improve both the effectiveness and efficiency of conventional classification systems such as multi-layer perceptrons and K-nearest neighbours by minimising redundant and noisy features. This is of particular significance for on-board image classification for future Mars rover missions. Left is an example MER image used as input to the FRFS-based classification software. Middle is the the resultant colour coded terrain classification. Right is the resultant segmented image. Link to International Journal of Hybrid Intelligent Systems 2011 paper here.

  

Aberystwyth PATLab and Local Field Trials

• The aim of the Planetary Analogue Terrain Laboratory (PATLab) is to allow comprehensive mission operations emulation experiments to be performed.
• The PATLab includes a 50 m² landscaped terrain region composed of Mars Soil Simulant-D (from DLR, Germany).
• The terrain includes an area for sub-surface sampling and a collection of fully characterised 'science target' rocks.
• The PATLab is heavily instrumented and its data and control facilities are available remotely via high-speed network links.
• EU FP7 funding for the Europlanet RI project has allowed the PATLab to become a TransNational Access Laboratory. EPSC 2008 abstract here (PDF).
• Extensive use is made of our half-size rover chassis which is based upon the ESA ExoMars rover Concept-E mechanics.
• The rover has 6-wheel drive, 6-wheel steering, and a 6-wheel walking capability (thus 3 DoF per wheel), and supports a panoramic camera instrument and a 3 DoF robot arm, in addition to onboard computing and communication facilities.
• To augment our PATLab based work, especially when testing our PanCam image processing work, we undertake field trials at the near-by Ynyslas and Clarach Bay beaches. The combination of sand dunes and sedimentary (sandstone and shale) rock faces provides an excellent environment for field trialing our work.

Past and Current Space Robotics Team Members (from 2002)

• Prof. Dave Barnes, Head of Space Robotics.
• Dr. Mark Neal, Co-I on EU FP7 PRoViScout project, Senior Lecturer in Computer Science.
• Dr. Fred Labrose, Co-I on EU FP7 PRoViScout project, Lecturer in Computer Science.
• Dr. Martin Wilding, Co-I on PanCam Calibration Target project, Senior Lecturer in IMAPS.
• Dr. Changjing Shang, Research Fellow in Space Robotics (previously RAE Daphne Jackson Research Fellow).
• Dr. Laurence Tyler, EU FP7 funded PDRA.
• Dr. Stephen Pugh, UK Space Agency funded PDRA.
• Dr. Richard Fallows, EU FP7 funded PDRA.
• Mr. Ralph Smithen, EU FP7 funded RA.
• Mr. Matt Gunn, Research Technician, IMAPS (part funded by UK Space Agency).
• Mr. Juan Fernando Vizcaya García, PhD Research Student.
• Mr. Rohit Jonahs, Department funded PhD Research Student.
• Mr. Chen Gui, Department funded PhD Research Student.
• Mr. Lilan Pan, Department funded PhD Research Student.
• Mr. John Pierson, NERC funded PhD Research Student.
• Dr. Roddy O'Hara, ESF funded Research Student.
• Dr. Claire Rocks, EPSRC funded PhD Research Student.
• Mr. Robin Xu Zheng, MPhil Research Student.
• Dr. Julian Henley, EPSRC ICASE funded PhD Research Student.
• Dr. Eddie Taylor, APRS funded PhD Research Student.
• Dr. Andy Shaw, ESA funded PDRA.
• Dr. Phil Summers, ESA funded PDRA.

Space Robotics Degree Schemes at Aberystwyth University

• The Department of Computer Science offers many undergraduate degree schemes that are very relevant to anyone wanting a career in Space Robotics.
• For postgraduate students Computer Science offers an MSc in Intelligent Autonomous Systems and an MSc in Intelligent Systems.
• The Insitute of Mathematics and Physics (IMAPS) offers a Space Science and Robotics degree scheme.
• For students wishing to study for a PhD degree in Space Robotics, application forms can be obtained from the Postgraduate Admissions Office.

For further details of the space robotics research at Aberystwyth University: email: dpb@aber.ac.uk

The information provided on this and other pages by me, Dave Barnes, is under my own personal responsibility and not that of Aberystwyth University. Similarly, any opinions expressed are my own and are in no way to be taken as those of Aberystwyth University.