Part 1: What is a Robotic Spacecraft?
In this first of three short papers, I introduce some of the basic concepts of space engineering with an emphasis on some specific challenging areas of research that are peculiar to the application of robotics to space development and exploration. The style of these short papers is pedagogical and this paper stresses the unique constraints that space application imposes. This first paper is thus a general introduction to the nature of spacecraft engineering and its application to robotic spacecraft. I consider the constraints and metrics used by spacecraft engineers in the design of spacecraft and how these constraints impose challenges to the roboticist. The following two papers consider specific robotics issues in more detail.
1. Introduction
The space environment represents one of the most challenging applications of robotics. Indeed, there is a widely-held but contentious viewpoint that space application represents a natural and inevitable arena for the advancement of robotics by imposing the requirement for high autonomy in space robotic systems. A minority extension of this viewpoint is that robotics is a discipline that has been stultified by its association with manufacturing, and space exploration provides an essential application in order to advance robotics as a discipline further towards its goal of developing human-like capabilities in the machine. Regardless of whether this may be so, or not, space application of robotics imposes unique drivers on robotics technology. The metric for success in space systems is the same as that for biological organisms – survival in a hostile and unrelenting environment. In this paper, I introduce some of the concepts of spacecraft engineering and how this impacts the design of robotic systems for space. I end the paper which some specific applications of robotics to space development and exploration to introduce two such applications that will be explored in the subsequent two papers.
2. Robotic Spacecraft
The first port of call in this paper is to put to rest a contentious, and often emotive, argument that plagues the political arena of space exploration. Every few years (the most recent following the Shuttle disaster) , the eternally-resurgent question of whether humans or robots should be adopted for space exploration is dusted off for regurgitation (Ellery, A., 2003). This debate is misplaced – there is a well-defined distribution of tasks across the human and the machine, and this distribution is of an evolutionary nature. There are tasks that are suited to robotics, and, likewise, there are tasks suited to humans. Robotics serves to ease the burden of more manual and repetitive tasks from the human astronaut allowing his/her deployment to tasks requiring the beyond the state-of-the-art machine intelligence. There is little doubt that human spaceflight provides a degree of flexibility in space activities that is unattainable in robotic missions.
As the capabilities of robotics become more sophisticated over time, so the role of humans will shift exclusively to tasks of greater complexity (Ellery, A., 2001). However, human exploration missions will always require prior reconnaissance by robotic missions – robots do not suffer the fragility of the human body and can reach further into outer space than human beings. Human space missions and robotic space missions are complementary. The human v robots debate is thus futile at best and vacuous at worst. There is little doubt that robotics and automation has great potential in space activities. It is uncontroversial that no space system in the foreseeable future will be entirely autonomous. However, the space engineering community have a particular aversion to placing their trust in machines, preferring to rely on the human being, be it ground operator or astronaut, to oversee, and often even manually control space activities such as rendezvous and docking. This emplacement of control on the human being can be dangerous – human performance is limited by strength, vigilance, fatigue and reaction speed. Indeed, human error has been the root cause of 65-70% of civil airline accidents. The general lesson is that if a procedure can be automated safely, then it should be. Automation is commonly adopted for fault diagnosis, power management and scheduling, and active thermal control of spacecraft.
The second port of call is to define terminology: “robotic spacecraft” is a generic term used to refer to deep space probes of all types with an emphasis on planetary explorers, but often used also to refer to space telescopes. The term emphasises their unmanned nature with the implication of significant degrees of autonomy, particularly for deep space probes that are characterised by deployment at great distances. In this paper, I shall describe the constraints on spacecraft design that impose stiff requirements on the implementation of robotics for space application.
3. Spacecraft Design
The first constraint imposed on the robotic spacecraft is the necessity of functioning in a hostile, non-terrestrial environment. All spacecraft must survive the stresses of launch, the vacuum and radiation of space, and for planetary deployment, the stresses of landing and the environment of the target planet. The application of robotics to spacecraft engineering imposes its own demands on the spacecraft engineer. All space missions are designed to achieve the mission goals that traditionally have been telecommunications provision, Earth observation (including meteorological) data return, military expediency, navigation functions, or scientific data return. Most spacecraft to date have thus been almost entirely designed as platforms for sensors for the collection of data and its transmission from space to Earth without physical interaction with the space environment (Shaw, G., Miller, D. & Hastings, D., 2000). Most spacecraft actuation mechanisms have been associated with propulsion, attitude control or mechanical deployment of large structures such as communications antennae and solar array panels. However, robotic actuation under closed loop control is becoming increasingly important for future space missions. The addition of robotic actuation imposes an order of magnitude increase in complexity to spacecraft design in terms of the performance of tasks that physically interact with their environments. For robotic space and planetary exploration, these environments may be partially or totally a priori unknown. Indeed, there is a peculiar contradiction between the spacecraft engineer who tends to avoid mechanical actuation systems as potential single point failure modes, and the roboticist for whom actuation provides the mode of interaction with the environment.
Spacecraft are designed according five main design budgets:
* cost budget which imposes a ceiling on the costs of the design, development, construction, validation, and launch of a spacecraft;
* mass budget which imposes a ceiling on the total mass of the spacecraft to be integrated into the launcher; severe limitation in mass - this favours lightweight designs using composite materials with consequent imposition of structural flexibility;
* propellant budget which imposes limits on the manoeuvring capability of the spacecraft once it is in orbit (this is a sensitive function of the total mass of the spacecraft);
* power budget which imposes a limitation of the power and energy available to each spacecraft subsystem and the payload - this favours the use of low power electronics, high efficiency motors, with high efficiency power scheduling with degraded computational resources;
* data budget which imposes severe limits on the capacity of the communications link to the ground and on the processing and storage capacity of onboard computing systems - shared human-machine autonomy is essential with sophisticated operator interfaces with predictive graphics.
All spacecraft are comprised of eight major subsystems:
* propulsion system which includes propellant, tankage, pumps, etc for manoeuvring in orbit;
* attitude control subsystem which control the orientation of the spacecraft to ensure that all components point in the correct direction, eg. solar panels point to the sun, thermal radiators to deep space, communications antennae to the Earth, and payload sensors to the targets of interest;
* structural and mechanical subsystem which mounts all other components onto the spacecraft in their required orientations, deploys all packaged components, and provides structural integrity to the spacecraft;
* power subsystem which generates, stores, distributes and controls all electric power to all the spacecraft components including the payload;
* thermal control subsystem which ensures that all components are maintained within their operational and/or survivable temperature ranges;
* communications subsystem which maintains the radio frequency communications link between the spacecraft and Earth to ensure uploading of commands and downloading of telemetry and payload data;
* onboard data handling (computer) subsystem which acquires, processes and stores all command, telemetry and payload data onboard the spacecraft;
* payload subsystem which provides the business end of the spacecraft, typically sensors and scientific experiments but may comprise a robotic system.
An additional constraint is the need for high reliability to ensure that the spacecraft survives until the end of its mission. This constraint places high emphasis on validation and testing of components and assemblies under space-like conditions. Of particular significance is the requirement for high payload capacity as a fraction of total mass as this defines the raison d’etre of the mission.
The payload may include a robotic system comprising:
* human-computer interfaces including telerobotic control and virtual reality interfaces;
* real-time control system including trajectory planning and generation, feedback servo-control laws, etc;
* onboard computer (probably a dedicated control computer) with fault tolerance;
* sensors and sensor processing, including stereo-vision, etc;
* actuation systems including electromechanical motors, etc.
All these subsystems must be integrated into a complete, functional, reliable spacecraft yet achieve this within the design budget constraints:
* complete system integration of large number of complex subsystems - this requires extensive testing and validation (under simulated thermal vacuum and launch conditions in particular) and tight project management;
* long lifetime (>10 y typically) during which the spacecraft must operate reliably without frequent maintenance with self-checking diagnostics but have the capacity for upgrade and repair-by-replacement of modules;
* extremely high reliability/safety (>90% typically) with fault tolerant (reliability) and triple fail-safe (safety) design protection - product assurance requires deterministic software approaches, disfavouring soft computing methods, and limitations in mechanical complexity, particularly moving parts which require special lubrication and/or hermetic sealing methods for operation in vacuum.
The implications to the roboticist to these constraints are thus: (Ellery, A., 2000; Putz 1998):
* the robotic spacecraft and its components must have lightweight construction to minimise its launch mass yet survive launch/impact loads (eg. up to 20 g axial acceleration and 145 dB acoustic noise for launch);
* the launch configuration is limited by the volume available in the payload fairing of the launcher necessitating that large area constructions be folded for launch and deployed reliably once in orbit (typically single-shot mechanisms);
* operate in vacuum environment (of 10 -3 Pa at LEO and 10 -15 Pa at GEO) - this requires the use of construction materials which are resistant to outgassing in vacuum, the use of dry lubrication in mechanisms, brushless electronic commutation in motors, and elimination of ultrasound as sensory modality;
* microgravity conditions has particularly ramifications for robotic dynamic control algorithms such that there is no ground reaction in space and non-linear dynamics effects become important, favouring low speed motion (~0.01m/s) as space lacks a damping medium for the dissipation of vibrations generated by the movement of boom-type configurations, and smooth motion profiles with high gearing ratios;
* the robotic spacecraft must endure severe temperature extremes and thermal cycling (-120 oC to +60 oC typically) - multilayer insulation, electric heaters, heat pipe and thermal radiators are required to maintain thermal limits;
* sensitive components such as electronics, computers, sensors and instrumentation must function under exposure to a high charged particle radiation environment (around 10 6 rad/y) - electronics and computing equipment require some shielding or hardening against SEUs; the limited capacity for onboard computational resources available (traditionally, radiation-hard processors are used, though there have been recent implementations of using COTS (commercial off-the-shelf) processors for small satellites in low Earth orbits) imposes severe restrictions on control and navigation algorithm complexity that can be implemented for real-time control;
* the lack of grounding of the robotic spacecraft may induce spacecraft electrostatic charging and discharges;
* the space environment provides a highly variable illumination environment with extreme constrasts and shadowing effects due to a lack of atmospheric scattering effects - this makes image processing difficult;
* the robotic spacecraft must operate for significant periods of time without direct human intervention (except for telecommanding and software uploading through the radio communications channel of limited bandwidth but these are subject to high signal transmission delays due to time-of-flight distance between the ground and space, and limited communication windows due to eclipsing disruption to the line-of-sight channel) - this imposes a need for significant levels of sensor-based autonomy with high fidelity ground operator interfaces.
4. Space Applications of Robotics
Although we have considered general robotic spacecraft issues here which are of critical importance to the space roboticist, space robotics as a discipline is focussed on more specific issues and reflects more closely the subject-area covered by terrestrial robotics. Indeed, space robotics, like its terrestrial counterpart, is generally divided into two subject-areas (though there is significant overlap):
* robotic manipulators – such devices are proposed for deployment in space or on planetary surfaces to emulate human manipulation capabilities; they may be deployed on free-flyer spacecraft or on-orbit servicing of other spacecraft, within space vehicles for payload tending, or on planetary landers or rovers for the acquisition of samples;
* robotic rovers – such devices are proposed for deployment on planetary surfaces to emulate human mobility capabilities; they are typically deployed on the surfaces of terrestrial planets, small bodies of the solar system, planetary atmospheres (aerobots), or for penetration of ice layers (cryobots) or liquid layers (hydrobots).
In the following two papers, I shall consider two case studies, one from each of these two topics: the use of manipulators mounted onto free-flying spacecraft to provide on-orbit servicing tasks, and planetary surface rovers for providing terrain-crossing mobility. I have specifically selected these two case studies to illustrate two issues – in the first, I consider the modification of traditional robotics techniques to the space environment; and in the second, I consider how new techniques may be borrowed from other disciplines (namely, vehicle terrainability) and applied to robotic planetary rovers.
5. References
Ellery, A. (2003). Human versus robots for space exploration and development. Space Policy 19, 87-91
Ellery, A. (2001). A robotics perspective on human spaceflight. Earth, Moon & Planets 87 (3), 173-190.
Ellery. A. (2000). An Introduction to Space Robotics, Springer-Praxis Publishers, , Chichester, UK.
Putz, P. (1998). Space robotics in Europe: a survey. Robotics & Autonomous Systems 23, 3-16.
Shaw, G., Miller, D. & Hastings, D. (2000). Generalised characteristics of communication, sensing and navigation satellite systems. J Spacecraft Rockets 37 (6), 801-811
Part 2: Space-based Manipulators
In this second of three short papers, I introduce some of the basic concepts of space robotics with an emphasis on some specific challenging areas of research that are peculiar to the application of robotics to space infrastructure development. The style of these short papers is pedagogical and the concepts in this paper are developed from fundamental manipulator robotics. This second paper considers the application of space manipulators to on-orbit servicing (OOS), an application which has considerable commercial application. I provide some background to the notion of robotic on-orbit servicing and explore how manipulator control algorithms may be modified to accommodate space manipulators which operate in the micro-gravity of space.
1. Introduction
Few engineered systems are expected to survive and function for more than a few years up to a decade or more without human intervention for servicing, maintenance or upgrading. Spacecraft are one of the few, engineered long-life products of human society that are denied such service and maintenance as part of their operational lifecycle. Simplistic notions of inaccessibility are no longer tenable as an excuse for this - the technology is in place to realise robotic on-orbit servicing (OOS). The traditional approach to spacecraft reliability has been through emphasis on high reliability, high cost components, and extensive validation and testing which also contribute to the expense of space platforms. The recent worrying trend in increasing on-orbit spacecraft failures has provided considerable support to the failure of this approach (Sullivan, B. & Akin, D., 2001). For any space platform, it is desirable to increase operational availability which requires a mixed approach:
Availability,
A = (MTBF/(MTBF+MTTR+MTFS)
where
MTBF=mean time between failures and reflects reliability
MTTR=mean time to repair and reflects maintainability
MTFS=mean time for supply and reflects logistic capability
Given the failure of reliability alone approaches to maximising spacecraft availability, maintenance of Earth-orbiting spacecraft through on-orbit servicing is essential by reducing MTTR and MTFS. Servicing of satellites may be implemented in all major orbits currently inhabited by Earth orbiting satellites. There are a number of families of orbits used by spacecraft today. Low Earth orbit (LEO) capped by the lowest point of the inner van Allen radiation belt at 2,000 km altitude is utilised by human missions and Earth observation missions (at polar inclinations). Medium Earth orbit (MEO) resides between the inner and outer Van Allen radiation belts centred around 10,000 km altitude and is ideal for mobile satellite constellations, eg. GPS constellation reside in 18,000 km altitude orbits. Most communications satellites reside in geosynchronous equatorial orbits (GEO) at 36,000 km altitude (though many Russian satellites utilise the high inclination Molniya orbits for access to high latitudes). In addition, there are highly elliptical orbits (HEO) that are used for some astronomy missions. For future astronomy missions, the Sun-Earth L 1 (for solar observations, eg. SOHO) and L 2 (for deep space observations, eg. Microwave Anisotropy Probe, Next Generation Space Telescope, Terrestrial Planet Finder) Lagrange points are expected to be the orbits of choice. On-orbit servicing has considerable potential for commercial applications in providing a space-based infrastructure (Ellery, A., Welch, C., & Curley, A., 2001). It has been suggested that the European Robotic Arm (ERA) on the ISS might be used to support astronomy missions by upgrading their instruments as an ISS-based servicer manipulator (Ellery, A. 2003).
2. On-Orbit Servicing
The Solar Maximum Repair Mission (SMRM) of 1984 was the first demonstration of on-orbit servicing by astronauts in combination with software workarounds uploaded from the ground, and teleoperation of the Shuttle Remote Manipulator System (SRMS) by an astronaut. The Solar Maximum Repair Mission represents a "textbook" case of OOS, involving the exchange of ORU (Orbital Replacement Unit) modules. Although the more complex tasks were performed by astronauts on EVA (extravehicular activity), such servicing operations may potentially be performed by robotic servicers. The repair and servicing of the Hubble Space Telescope and other US astronaut activities have further demonstrated the feasibility of space-based servicing tasks. Indeed, robotic servicing was an instrumental part of the early stages of the ISS programme in which two concepts were proposed to perform these functions - the Flight Telerobotic Servicer (FTS) and the Orbital Maneouvring Vehicle (OMV) - but these were cancelled in the face of budget cuts. NASA's AERCam (Autonomous Extravehicular Robotic CAMera) represents a step in this direction – AERCam is a small 35 cm diameter freeflying sphere comprising a camera for aiding astronaut EVA, thrusters for attitude and translation control, and avionics developed from astronaut MMU (manned maneouvering unit) technology. The addition of robotic manipulators onto a larger spacecraft platform would offer freeflyer servicer capabilities. The sizing of the manipulator would be determined by EVA-equivalence, one example of which is the proposed ESA dextrous robot manipulator system:
1. Seven degrees of freedom (three degrees of freedom at the shoulder, one degree of freedom at the elbow, and three degrees of freedom at the spherical wrist)
2. Three fingered end effector with cylindrical dimensions 10 x 15 cm
3. Control set-point rate of 100 Hz
4. Forward reach of 1m - this requires multiple grappling points on the target as full reachability around most satellites would require a reach of 4.5 - 16 m which is impractical
5. End effector position accuracy of 0.3 mm/0.1 o
6. Maximum end effector velocity of 0.1 m/s and 0.1 rad/s
7. Structural displacement compliance of 1x10 6 N/m and rotational compliance of 5x10 4 Nm/rad
8. Force/torque exertion of 200 N and 20 Nm respectively
9. Payload capacity of 500 kg in microgravity environment
Fig. 1. ATLAS robotic servicer concept Fig. 2. Single arm version of the robotic servicer kinematics and dynamics (courtesy Praxis Publishers)
The Japanese ETS (Engineering Test Satellite) VII of 1996 has demonstrated many of the basic technologies for robotic on-orbit servicing, including autonomous rendezvous and docking to a target, teleoperated manipulator movement control whilst maintaining stable attitude, and vision/force feedback based peg-in-hole tasks. All robotic servicers will be required to grapple the target spacecraft for retrieval, resupply of consumables, repair, or retrofit. In the US, one of the most advanced servicer concepts is Ranger for which a neutral buoyancy test vehicle has been developed. One proposal for such a robotic servicer is 1.5 tonne ATLAS (Advanced TeleRobotic Actuation System) – see Fig. 1.:
3. Control of Space Manipulators
On-orbit servicing robotics is a modern version of an old field that stems back to the origins of science itself – Newton (1642-1729), Euler (1707-1783), d’Alembert (1717-1783), Lagrange (1736-1813), Laplace (1749-1827) and Hamilton (1805-1865) all contributed to the development of mechanics and dynamics. The primary differentiating characteristic of on-orbit servicing robotics from terrestrial robotics is that the robotic servicer operates in microgravity. Whereas the terrestrial manipulator is mounted onto terra firma, in space there is no such reaction force and torque cancellation – the motion of the manipulator(s) will generate reaction forces and moments on the spacecraft at the manipulator mounting point(s). Robotic freeflyer manipulators are difficult to control as the spacecraft platform moves in response to the manipulator movements. A free-floating platform no longer permits the use of the base of the manipulator as the inertial coordinate frame of reference. If this effect is not taken into account, the manipulators will overshoot the target that it is attempting to grapple. A similar effect occurs with astronauts in the microgravity environment of space. They undergo changes in psychomotor performance and posture and their limb movements tend to overshoot their targets until the astronaut's brain has adapted to the new microgravity conditions (normally within two to three days). The robotic manipulator control system must similarly compensate for operating in microgravity while implementing the computationally intensive algorithms for trajectory interpolation, inverse kinematics, dynamics, and force/position control of the end effector. We may apply the conservation of momentum to the freeflyer servicer system (assuming a single manipulator for simplicity) in order to apply constraints to solve the problem, which allows us to define the centre of mass of the whole system to lie at the origin of the inertial reference frame - see Fig. 2.:
The position of the manipulator end effector with respect to inertial space may be represented by:
Although a number of global dynamics techniques have been proposed, they suffer from high computational complexity problems (Umetani, Y. & Yoshida, K., 1989). Now, although conservation of linear momentum is intregrable to yield constraints on linear position of the end effector, this is not the case for angular momentum conservation, which is not integrable to unique angular constraints as it is a non-holonomic constraint. It is possible to separate out the rotational and translation components of the system to yield a much simpler and more readily implementable set of control algorithms. We employ a dedicated spacecraft attitude control to which a feedforward signal from the robotic manipulator system may be computed as a by-product of the Newton-Euler dynamics formulation of the manipulator (Longman, R., Lindberg, R., & Zedd, M., 1987) which computes the reaction moment exerted at the manipulator mount point on the spacecraft as:
where
The values of F T and N T are computed as a by-product of the Newton-Euler dynamics formulation for the manipulator to ensure that R 0 = I 3. This provides the basis for stabilisation of the attitude of the spacecraft platform. The translation effect needs to be compensated for, and this can be done through a variation on the terrestrial Denavit-Hartenburg matrix formulation thus (Ellery, A., 2004b):
where R=3x3 direction cosines matrix as for terrestrial manipulators
= inertial position of the end effector
= kinematic-dynamic parameter of each link i
Similarly, the Jacobian may be given by providing the basis for resolved motion control algorithms (Ellery, A., 2004b) such as the computed torque controller and force control algorithms such as the hybrid position/force controller. This is applicable to any geometry of manipulator of n degrees of freedom, which is determined by the four Denavit-Hartenburg parameters. These results mean that terrestrial robotic control algorithms may be used with only minor modifications for the control of a space manipulator, easing the computational burden on space-rated processors for real-time control.
4. Implications
I have found that use of the above algorithm suggests that realistic servicer designs such as ATLAS require the use of control moment gyroscopes for spacecraft attitude control, particularly when implementing force control (Ellery, A., 2004a). The formulation presented above is readily extended to two or more manipulators allowing the implementation of onboard closed loop control of space manipulators mounted onto robotic servicer spacecraft. Robotic on-orbit servicing of spacecraft is achievable today and indeed, a number of space agencies are currently investigating this possibility in the near future. The opportunity for OOS has important implications for the development of the space environment – in providing a fundamental part of space infrastructure, OOS represents the first tentative steps towards the development of a fully functional space-based manufacturing capability with material processing and assembly, and only when this is achieved, will space become a part of the everyday world.
5. References
Ellery, A. (2003). Robotics in the service of astronomy. Astronomy & Geophysics 44 (3), 3.23-3.25
Ellery, A. (2004a). Robotic in-orbit servicers – the need for control moment gyroscopes for attitude control. In press with Aeronautical J
Ellery, A. (2004b). An engineering approach to the dynamic control of space robotic on-orbit servicers. In press with Proc Inst Mech Eng, Part G: J Aerospace Eng .
Ellery, A., Welch, C. & Curley, A. (2001). A proposed public-private partnership for the funding of robotic in-orbit servicers. Space & Robotics Conference , March 2001, Albuqerque, N Mexico, USA.
Ellery. A. (2000). An Introduction to Space Robotics, Springer-Praxis Publishers, , Chichester, UK
Longman, R., Lindberg, R., & Zedd, M. (1987). Satellite-mounted robot manipulators – new kinematics and reaction compensation. International J Robotics Research 6 (3), 87-103.
Sullivan, B. & Akin, D. (2001). Survey of serviceable spacecraft failures. AIAA 2001-4540
Umetani, Y. & Yoshida, K. (1989). Resolved motion rate control of space manipulators using a generalised Jacobian matrix. IEEE Trans Robotics & Automation 5 (3), 303-314.
Space Robotics
Alex Ellery
Surrey Space Centre, University of Surrey, Guildford, Surrey, United Kingdom Contact:
Published in:
Part 1: Volume 1, Number 2, June.2004
Part 2: Volume 1, Number 3, September 2004
Part 3: Volume 1, Number 4, December 2004
For further informations in German please click here!