Deep space exploration missions have received a lot of attention in the world, such as the moon, Mars, asteroids for scientific observation and human activities. Recently, not only space agencies but also industries or universities have earnestly studied and developed lunar or planetary robots for lunar or planetary exploration. International cooperative missions are also under discussion including human lunar exploration. Space robotics including Al is a key technology for planetary exploration as well as orbital service. Future space programs require space AI and robotics technology to construct, repair and maintain satellites and space structures in orbit. This paper introduces recent space missions and surveys hot topics on new research activities and developed technology for the near future missions.

In this paper, we describe an ongoing multi-institution study in using emplaced computational resources such as high-volume storage and fast processing to enable instruments to gather and store much more data than would normally be possible, even if it cannot be downlinked to Earth in any reasonable time. The primary focus of the study is designing science pipelines for on-site summarization, archival for future downlink, and multisensor fusion. A secondary focus is on providing support for increasingly autonomous systems, including mapping, planning, and multi-platform collaboration. Key to both of these concepts is treating the spacecraft not as an autonomous agent but as an interactive batch processor, which allows us to avoid “quantum leaps” in machine intelligence required to realize the concepts. Our goal is to discuss preliminary results and technical directions for the community, and identify promising new opportunities for multi-sensor fusion with the help of planetary researchers.

S PACEmissions are faced with numerous challenges, in which the thoughtful application of automation and autonomy and the integration of artificial intelligence (AI) techniques can contribute significantly and directly to mission success. Among these challenges are operating in uncertain and extreme environments, managing scarce resources such as power, communications, and computation under severe constraints, and grappling with light-time delays, as well as accomplishing unprecedented functionality such as precision landing on planetary surfaces and achieving science return supportive of discovery, all thewhile grapplingwith the intricate complexity of systems engineering tied to requirements for reliability, robustness, and safety. This Special Issue of the Journal for Aerospace Information Systems pertains to intelligent systems for space exploration. Its publication was inspired by the International Symposium for Artificial Intelligence, Robotics, and Automation in Space (iSAIRAS) conference series. However, there was no requirement that work collected in this issue be derived specifically from reports at an iSAIRAS conference. As guest editors, we solicited work across the following scope of interest: 1) space-based demonstration or application of intelligent systems concepts, 2) ground-based demonstration of autonomous space systems concepts, 3) ground-based demonstration or application of mission operations automation, and 4) laboratory demonstration of AI-based concepts for space missions. Within this general scope, relevant technical topics included space systems autonomy (onboard software for mission planning and execution; resource management; fault protection; science data analysis; guidance, navigation, and control; smart sensors; testing and validation; and architectures) and mission operations automation (decision support tools for mission planning and scheduling, anomaly detection and fault analysis, innovative operations concepts, data visualization, secure commanding and networking, and human–robotic teaming). Submitters were asked to provide technical descriptions of systems and results and analysis of experimentation. Lessons learned in development and operations were also solicited, as relates to systems engineering, testing, and validation. As a final consideration of scope, we encouraged papers addressing any operating regime for space exploration from Earth orbit to deep space (planetary and small-body orbital environments and surfaces), including both robotic and human–robotic mission concepts. In the end, it is clear that there is an abundance of quality research being conducted internationally within the identified scope. The five articles appearing here represent that scope as aworthy sample, but do not come close to exhausting the relevantwork of high quality that is underway. The following is a summary of the specific content of this Special Issue. An implicit but important challenge in achieving success with autonomy, beyond the critical-path research challenges, involves working with end users and stakeholders tomake the case for the added value andmanaged risk of the new capability, often in the face of initial skepticism. Burl et al. describe not only their technical success to develop an autonomous vision-based capability to detect scientifically interesting phenomena on the surface of Mars, but also provide details of what it takes to successfully deploy such an unprecedented capability. An architecture that has proven useful for supporting autonomous science operations is one that enables event detection and response. Chien et al. report on just such an architecture, hosting capabilities for classifying content and determining salient features within images collected by an Earth observing hyperspectral instrument, integrated with an onboard (re)planning capability for scheduling additional observations, while respecting resource and other operating constraints, all demonstrated on a CubeSat platform. As future space exploration grows in scale, missions will increase in scope and deploy ever greater numbers of assets to other bodies for extended missions of years or more. In this evolution, missions will need robotic explorers to act with increasing autonomy. Yliniemi et al. highlight work in developingmulti-agent learning systems that present a promising approach to addressing learning andmulti-agent coordination for such future missions. A basicmotivation for provisioning a platformwith autonomy capability is to have an effectivemeans of grapplingwith operational uncertainty in a remote environment. Burroughes et al. describe an architectural approach and formalism for responding to the inevitable unforeseen events through informed reconfiguration of the system. They assess the computational efficiency of their approach through experimentation. One persistent challenge in space exploration is onboard data management. Spacecraft have ever-increasing capabilities to acquire large amounts of science data, but limited ability to store and downlink such data. This challenge is complicated by the prevalence of content-dependent compression schemes, which make prediction of the amount of data acquired uncertain. Maillard et al. present an approach that involves both the flight and ground system to adaptively manage the data acquired, stored, and downlinked to optimize spacecraft operations. TheGuest Editors hope that readers will find these articles edifying as to current research topics, as well as state-of-the-art autonomy capability for space systems. The Guest Editors also wish to thank all authors who submitted articles and the set of capable reviewers.

Spaceflight computing is a key resource in NASA space missions and a core determining factor of spacecraft capability, with ripple effects throughout the spacecraft, end-to-end system, and mission. Onboard computing can be aptly viewed as a "technology multiplier" in that advances provide direct dramatic improvements in flight functions and capabilities across the NASA mission classes, and enable new flight capabilities and mission scenarios, increasing science and exploration return. Space-qualified computing technology, however, has not advanced significantly in well over ten years and the current state of the practice fails to meet the near- to mid-term needs of NASA missions. Recognizing this gap, the NASA Game Changing Development Program (GCDP), under the auspices of the NASA Space Technology Mission Directorate, commissioned a study on space-based computing needs, looking out 15-20 years. The study resulted in a recommendation to pursue high-performance spaceflight computing (HPSC) for next-generation missions, and a decision to partner with the Air Force Research Lab (AFRL) in this development.

Coming years will bring several comet rendezvous missions. The Rosetta spacecraft arrives at Comet 67P/Churyumov-Gerasimenko in 2014. Subsequent rendezvous might include a mission such as the proposed Comet Hopper with multiple surface landings, as well as Comet Nucleus Sample Return (CNSR) and Coma Rendezvous and Sample Return (CRSR). These encounters will begin to shed light on a population that, despite several previous flybys, remains mysterious and poorly understood. Scientists still have little direct knowledge of interactions between the nucleus and coma, their variation across different comets or their evolution over time. Activity may change on short timescales so it is challenging to characterize with scripted data acquisition. Here we investigate automatic onboard image analysis that could act faster than round-trip light time to capture unexpected outbursts and plume activity. We describe one edge-based method for detect comet nuclei and plumes, and test the approach on an existing catalog of comet images. Finally, we quantify benefits to specific measurement objectives by simulating a basic plume monitoring campaign.

Recent commercial developments in multicore processors (e.g. Tilera, Clearspeed, HyperX) have provided an option for high performance embedded computing that rivals the performance attainable with FPGA-based reconfigurable computing architectures. Furthermore, these processors offer more straightforward and streamlined application development by allowing the use of conventional programming languages and software tools in lieu of hardware design languages such as VHDL and Verilog. With these advantages, multicore processors can significantly enhance the capabilities of future robotic space missions. This paper will discuss these benefits, along with onboard processing applications where multicore processing can offer advantages over existing or competing approaches. This paper will also discuss the key artchitecural features of current commercial multicore processors. In comparison to the current art, the features and advancements necessary for spaceflight multicore processors will be identified. These include power reduction, radiation hardening, inherent fault tolerance, and support for common spacecraft bus interfaces. Lastly, this paper will explore how multicore processors might evolve with advances in electronics technology and how avionics architectures might evolve once multicore processors are inserted into NASA robotic spacecraft.

We investigate the possibility of forming deeply bound ultracold RbCs molecules by a two-color photoassociation experiment. We compare the results with those for Rb-2 in order to understand the characteristic differences between heteronuclear and homonuclear molecules. The major differences arise from the different long-range potential for excited states. Ultracold Rb-85 and Cs-133 atoms colliding on the X-1 Sigma(+) potential curve are initially photoassociated to form excited RbCs molecules in the region below the Rb(5S) + Cs(6P(1/2)) asymptote. We explore the nature of the Omega = 0(+) levels in this region, which have mixed A(1)Sigma(+) and b (3)Pi character. We then study the quantum dynamics of RbCs by a time-dependent wavepacket (TDWP) approach. A wavepacket is formed by exciting a few vibronic levels and is allowed to propagate on the coupled electronic potential energy curves. We calculate the time dependence of the overlap between the wavepacket and ground-state vibrational levels. For a detuning of 7.5 cm(-1) from the atomic line, the wavepacket for RbCs reaches the short-range region in about 13 ps, which is significantly faster than for the homonuclear Rb-2 system; this is mostly because of the absence of an R-3 long-range tail in the excited-state potential curves for heteronuclear systems. We give a simple semiclassical formula that relates the time taken to the long-range potential parameters. For RbCs, in contrast to Rb-2, the excited-state wavepacket shows a substantial peak in singlet density near the inner turning point, and this produces a significant probability of de-excitation to form ground-state molecules bound by up to 1500 cm(-1). The short-range peak depends strongly on non-adiabatic coupling and is reduced if the strength of the spin-orbit coupling is increased. Our analysis of the role of spin-orbit coupling concerns the character of the mixed states in general and is important for both photoassociation and stimulated Raman de-excitation.

The use of discrete variable representations is now commonplace in chemical dynamics calculations. In this paper, we employ spectral difference methods to speed up these calculations. We present five new spectral difference weight functions and compare them with those that already exist in the literature for two different bound state problems. We find that one particular weight we propose, based on a Gaussian function, outperforms all other weights.