A Alysis I Robotic Auto Omy for Future Pla Etary Missio S

Future planetary surface exploration missions as envisioned by international agencies would require of human and robot cooperation. Past, ongoing and upcoming surface robotics missions will pave the way for technology infusion and to demonstrate its readiness for critical-safety missions. Rovers with major autonomous capabilities are a must and are under investigation and development; however, there are some aspects highlighted in this article that cannot be operationally demonstrated with precursor missions by its nature. First, rovers that would require of manifold control modes from fully manual teleoperation to autonomous, where we propose to approach this flexible system architecture by defining different autonomy control levels. Second, rovers might be differently operated depending on the mission’s phase, where we address robotic taxonomies as a driver for designing operational needs. To initiate this topic, we summarise the state-of-the-art of autonomous capabilities of Mars Exploration Rovers gathered from NASA literature. 1 WHAT IS COMI G EXT? The human exploration of planets is on the roadmap of future missions as well as the fostering of international cooperation among agencies [1] and its proper coordination at European level [2]. The European Space Agency (ESA) has envisioned to explore the surface of Mars with humans by the year 2033 within the Aurora Programme, and the National Aeronautics and Space Administration (NASA) has targeted to explore the surface of Moon with humans by the year of 2020 within The Vision for Space Exploration. NASA plans to build bases on the lunar surface, where astronauts will live in pressurized habitats for a certain period of time to achieve the mission’s scientific goals. The scenario of such a mission shall include many elements, including Earth-Moon communication relay satellites, ground communication antennas, lunar bases, astronauts performing Extravehicular Activity (EVA) or Intra-vehicular Activity (IVA), astronauts travelling in pressurised vehicles from remote bases, robots to perform repetitive and hazardous tasks, instruments deployed on the lunar surface, and so forth. Figure 1 – Operational infrastructure for lunar exploration [3] Figure 1 illustrates the use of robotics within such a complex mission: ranging from autonomous rovers for transportation and logistics support capable of long traverses, dexterous manipulators in support of construction and assembly, robots to help astronauts in performing activities or performing surveillance, and for deployment and maintenance of surface assets, etc. There is a long journey before human and robots can explore with permanent bases on the Moon. First, all related infrastructure to support such mission must be developed: redundant launcher technologies, Earth Moon Lagrange (EML) communications relay points, Low Earth Orbit (LEO) or Low Lunar Orbit (LLO) stations, Lunar Descent Module (LDM) and so on [4,5,6,7]. All these technologies and infrastructures that would provide and ensure a sustainable Lunar or Martian surface exploration by humans, with the support of robots, will need an iterative development of key technologies and precursor missions to demonstrate and prepare the environment, as proposed by Christie et al. [8]: 1. Reconnaissance Rovers 2. Site Preparation and Simple Instalments 3. Permanent Moon Base Construction 4. Sustaining Lunar Infrastructure and Exploration Reconnaissance rovers are necessary before returning to the lunar surface to enhance the knowledge of the landing site (assumed to be polar, based on the current mission roadmaps) and its actual surface environmental conditions; to perform topology and general mapping of the lunar surface providing complementary data of precursor orbital missions, as the Japanese SELENE. The technology basis of these rovers would be the current state of the art, i.e. those deployed on Mars: the Mars Exploration Rovers (MERs), Mars Science Laboratory (MSL), ExoMars or the Phoenix Lander. The objective of these rovers would involve conducting surveying activities, geological analysis and deployment of initial navigation beacons and communication infrastructures. They would need to be capable of operating in a minimum of two modes: fully autonomous and remotely operated, initially from Earth-based control centres and eventually from Lunar-based stations. They would need direct-to-Earth communication capabilities yet also be capable of interfacing with both the beacons as well as any precursor lunar landers. Technologically, the rover would need surface feature recognition capabilities or some other means for relative localisation. Navigation solutions can be divided into relative localisation (e.g. LIDAR as primary and normal optics as secondary due to lighting lunar conditions), absolute heading (e.g. star tracker) and global localisation (e.g. radio localisation/descent imagery). Scientific and exploration tools and payloads would be controlled by pre-determined procedures or, as on the MERs, be controlled via remote control by scientists and operators on Earth. These rovers may also be equipped with secondary purpose assets in order to perform auxiliary experiments. The rovers might be also controlled by astronauts once humans arrive in manned missions, and need to be designed with enough endurance. Site preparation and simple instalments would include manned missions that shall benefit from more comprehensive survey datasets of the local lunar surface, with the aid of navigation and communication assets deployed on the surface. It is probable this phase involves the establishment of a lunar-based communications and robotics control centre, integrated with, or independent from, a habitat for astronauts. This centre might be equipped with all of the functionality of the terrestrial control centre with respect to robotic control, providing to the astronauts the capability to operate and coordinate missions on-site. This centre would interface with the terrestrial control centre such that critical mission data (asset location, telemetry data and video) could be shared. It may be possible that all such lunar assets would be supported by a lunar space station, located at LLO or EML points [7]. The LLO option would provide safe haven capabilities, lunar outpost logistics preparation and handling, support of cargo staging packages prior to shipment to lunar surface, communications enhancements and redundancy, as well as navigation support to lunar-based rovers. It shall be also capable of controlling any robotic assets on the surface freeing up of tasks to astronauts on the surface. The design basis for this LLO station would be the approach of the International Space Station (ISS) and its associated robotic elements, i.e. a set of highly specialised modules with standardised interfaces for docking, electrical and fluid connections, etc. This phase would also include delivery, set-up and test of heavy duty construction equipment using an autonomouslycontrolled robotic platform, upon which a variety of mission-specific modules could be mounted. This robot would be supervised by Moon-based astronauts. To complete the phase, a lunar personnel manned roving vehicle might be used to provide astronauts with a means of transportation, having the same operational characteristics as precursor rovers, with similar mobility capacities (terrain assessment, path planning, localisation and navigation) and communications interfaces. Permanent moon base construction shall build a series of interconnected standalone modules shipped sequentially; and implicate the crewmembers in various tasks, e.g. construction tasks like unpacking, checkout, transport, anchoring and connection to power grid and data lines, all with the help of surface rovers. Sustaining lunar infrastructure and exploration shall include the tasks of maintaining the outpost, although a more comprehensive study of the actual effects on long-term exposure to the lunar environment would be needed. Activities might include the installation of newer equipment once their useful lifetime is expired, routine tasks of supervising the equipment via remote-controlled or autonomous robots in order to reduce the astronauts’ exposure and safety of EVA. 2 OVERVIEW OF AUTO OMOUS CAPABILITIES MERs, MSL and ExoMars shall provide the technological heritage to future reconnaissance rovers as introduced previously. At the moment, in order to overview the state-of-the-art in autonomous capabilities from an operational perspective, we need to focus on the excellent performance that MERs have offered during their lifetime. The following table aims at giving an overall picture of onboard autonomous capabilities of the rovers, mostly focusing on mobility and autonomous science aspects [9], dated from 2007. We have also mapped the autonomy capability to an Autonomy Maturity Level (AML) recalling the operational experience and confidence during operations: AML Description Corresponding TRL 1 Technology under development for its use in space 3 5 2 Technology developed, tested, verified and validated in rover prototypes for its use in space 5-7 3 Technology experimentally tested during the mission in highly controlled conditions 7-8 4 Technology used during nominal operations over the mission 9 5 Technology extensively used during nominal operations and/or in contingency procedures 9 Table 1 – Definition of Autonomy Maturity Levels and comparison with Technology Readiness Level Capability Requirement Source Approach Sensors/SW AML Absolute Orientation Sensing (OBS) Onboard position estimate can accumulate several degrees of drift after integrating the gyros for thousands of seconds. Sun vector recalibration (and position) by pointing the camera where the Sun is supposed to be, and processing the image to reallocate its centre (or conventional sun sensor for lunar applications). PANCAM, Local Solar Time, Inertial Measurement Units (IMUs) 5 (used in Sojourner mission and extensively in MERs) Stereo Imaging Processing (SIP) Need for depth, geometry and shape information of surrounding terrain. Generate 3D measurements (disparity) of points in stereo images, performing a windowed 1D search using the sum-of-absolute-differences metric. Software relies on the camera's geometric lens calibration. PANCAM (2 stereopairs) and NAVCAMs with different FOV 5 (used in Sojourner mission and extensively in MERs) Local Path Selection (LPS) Detect drift conditions when traversing, e.g. in sandy slopes. Add attitude drift information processed from IMUs [algorithm Ali et al., 2005, 9] Wheel encoders and gyros (IMUs) 4 Visual Odometry (VO) LPS does not detect slippage conditions when traversing, e.g. in steep hillsides, mixed sand/rock terrains inside craters or sandy ripples in flat plains of Meridiani. Software compares pairs of NAVCAM images of nearby terrain to autonomously detect and track features between them. 2D and 3D motion of those features is used to update the onboard position estimate. [algorithm Maimone et al., 2007, 9] NAVCAM 4 Terrain Assessment (TA) Detect geometric hazards around the area to assist drive modes, e.g. rocks, ditches or cliffs. Clouds of 3D points are fitted in rover-sized patches of data to a plane. Software searches for 1) steep obstacles large deltas in elevation of best fit plane, 2) tilt hazards large angle between surface normal and the Up vector, 3) Roughness hazards residual of the planar fit. Software performs traversability analysis with up to 10 separate points of clouds; normally performed with a single stereo pair. SIP imagery Software tool: GridBased Estimation of Surface Traversability Applied to Local Terrain (GESTALT) 4