Progressive autonomy: a method for gradually introducing autonomy into space missions

This paper presents a method under development for introducing autonomy and agent-based software into future space- and ground-based missions while both reducing the risk of mission failures and gaining the confidence and support of mission management and principal investigators (PIs). This is being done using a mechanism to support dynamic agent-community evolution (e.g., agents adapting to community changes, agents joining a community, or agents leaving a community). This dynamic capability of agents is necessary to achieve what we call ‘‘progressive autonomy,’’ which will allow dynamic modification of satellite systems using agent migration to update and modify spacecraft capabilities on an as-needed basis, as well as allow the introduction of mission management and autonomy into existing missions. This paper will also address an application of progressive autonomy through spectral analysis automation (SAA). The long-term fully realized SAA system will be a multiagent system designed to provide automated support for two major functions: (1) the automatic remote filtering (onboard a spacecraft or robotic device) of spectral image data based on PI guidance, goals, and science agenda and (2) the packing and transmission of the selected spectral data to the PI for further processing. Additionally, the innovative multiagent-based infrastructure for the SAA can be generalized in such a way as to enable it to support the type of progressive autonomy that will be needed to support an adaptive and growing autonomous behavior for other spacecraft or robotic subsystems.     

On-board autonomy, EURECA experience and requirements for future space missions

The paper describes, after a basic definition of on-board autonomy, the on-board autonomous mission control capabilities used during the EURECA mission for nominal operations and fault management and related experiences. It also covers autonomy enhancements required for future highly productive missions. These are functionalities in relation to on-board mission planning, use of intelligent sub-system elements, “off-line” test capabilities and adaptive fault management.

EURECA, an unmanned retrievable multi-disciplinary science laboratory has successfully flown an 11-month mission in a 500 km, 28 deg. inclination orbit in the period August 1992 to June 1993. It was launched and brought back to earth by Space Shuttle flights STS46 and STS57 respectively. The complex spacecraft was operated by ESOC, in continuous productive mode, with less than 6% ground on-line contact, in a highly on-board autonomous mode. Its mission products encompassed all types of continuously produced scientific data, processed materials and biological samples, and a significant number of technological demonstrations.     

Miniature robots for space and military missions

Miniature robots enable low-cost planetary surface exploration missions as well as new military missions in urban terrain where small robots provide critical assistance to human operations. These space and military missions have many similar technological challenges. This article describes the technologies under development, the applications where these technologies are relevant to both space and military missions, and the status of the most recent technology demonstrations in terrestrial scenarios     

Intelligent control of life support for space missions

Future manned space operations will include a greater use of automation than we currently see. For example, semiautonomous robots and software agents will perform difficult tasks while operating unattended most of the time. As these automated agents become more prevalent, human contact with them will occur more often and become more routine, so designing these automated agents according to the principles of human-centered computing is important. We describe two cases of semiautonomous control software developed and fielded in test environments at the NASA Johnson Space Center. This software operated continuously at the JSC and interacted closely with humans for months at a time.     

Autonomous,biologically inspired systems for future space missions

Future space missions will go beyond reconnaissance, enabling sustained in-situ scientific studies. Such missions require space platforms that operate in a closed-loop fashion in its environment. The platforms must be autonomous, evolvable, resilient and highly distributed. The present article describes these characteristics, along with the mission concepts considered by NASA that require these characteristics. Autonomy is a practical application of AI. Evolvability and highly distributed characteristics are biologically inspired.     

Model-based tracking by classification in a tiny discrete pose space

A method is presented for tracking 3D objects as they transform rigidly in space within a sparse range image sequence. The method operates in discrete space and exploits the coherence across image frames that results from the relationship between known bounds on the object's velocity and the sensor frame rate. These motion bounds allow the interframe transformation space to be reduced to a reasonable and indeed tiny size, comprising only tens or hundreds of possible states. The tracking problem is in this way cast into a classification framework, effectively trading off localization precision for runtime efficiency and robustness. The method has been implemented and tested extensively on a variety of freeform objects within a sparse range data stream comprising only a few hundred points per image. It has been shown to compare favorably against continuous domain iterative closest point (ICP) tracking methods, performing both more efficiently and more robustly. A hybrid method has also been implemented that executes a small number of ICP iterations following the initial discrete classification phase. This hybrid method is both more efficient than the ICP alone and more robust than either the discrete classification method or the ICP separately     

Model-based space planning for temporary structures using simulation-based multi-objective programming

Construction trades need to share temporary structures to increase the output of direct work while controlling the labor input of indirect work. The purpose of this research is to develop a framework to determine the optimal location of temporary structures in a computerized practical manner for piping construction projects. Based on the spatial relationship between work envelope and scaffolding placement requirements, this paper presents the optimization model in two phases: the simulation-based optimization model and a multi-attribute utility (MAU) based alternative selection model. A multi-objective optimization model is established to improve scaffolding availability among multiple activities while maximizing piping crew productivity. The multi-attribute utility model is employed to handle the uncertainty of the assessment weights on the attributes to illustrate the preference of decision makers among different scaffolding placement alternatives obtained from the first phase. The approach was validated in a piping module, which provided superintendents and space planners with an effective decision-making tool among possible scaffolding alternatives in piping construction. The proposed optimization technique is an alternative methodology for solving the productivity-tasks-scaffolding trade-off problem, which further revolutionizes the spatial coordination process of workspace management and temporary structure planning.     

Model-based multi-objective decision making under deep uncertainty from a multi-method design lens

Several approaches within the exploratory modelling literature—each with strengths and limitations—have been introduced to address the complexity and uncertainty of decision problems. Recent model-based approaches for decision making emphasise the advantage of mixing approaches from different areas in leveraging the strengths of each. This article shows how a multi-method lens to the design of decision-making approaches can better address different characteristics of multi-objective decision problems under deep uncertainty. The article focuses on interactions between two broad areas in model-based decision making: exploratory modelling and multi-objective optimisation. The article reviews this literature using a specific multi-method lens to analyse previous researches and to identify the knowledge gap. The article then addresses this gap by demonstrating a multi-method approach for designing adaptive robust solutions. The suggested approach uses a Pareto optimal search from multi-objective optimisation for enumerating alternative solutions. It also uses Robust Decision Making and Dynamic Adaptive Policy Pathways approaches from exploratory modelling for analysing the robustness of enumerated solutions in the face of many future scenarios. A hypothetical case study is used to illustrate how the approach can be applied. The article concludes that a new lens from a multi-method design perspective is needed on exploratory modelling to provide practical guidance into how to combine exploratory modelling techniques, to shed light on exiting knowledge gaps and to open up a range of potential combinations of exiting approaches for leveraging the strengths of each.     

Payload‐centric autonomy for in‐space robotic assembly of modular space structures

The paper addresses the problem of constructing large space structures (～100 m) by using autonomous robots to assemble modular components in space. We are motivated by the problem of creating space structures at a scale greater than what is feasible with a single self‐deploying design. We had two goals in this work. The first was to investigate and demonstrate the feasibility of long‐order multitask autonomy. The second was to study the balance between required tolerances in hardware design and robotic autonomy. This paper reports on a payload‐centric autonomy paradigm and presents results from laboratory demonstrations of automated assembly of structures using a multilimbed robotic platform. We present results with deployable 20 lb payloads (1 m trusses) that are robotically assembled to form a 3‐m diameter kinematically closed loop structure to subcentimeter accuracy. The robot uses its limbs to deploy the stowed modular structural components, manipulate them in free space, and assemble them via dual‐arm force control. We report on results and lessons learned from multiple successful end‐to‐end in‐lab demonstrations of autonomous truss assembly with JPL's RoboSimian robot originally developed for the Defense Advanced Research Projects Agency (DARPA). Videos of these demonstrations can be seen at https://goo.gl/muNLJp (JPL, 2017 ). Each end‐to‐end run took precisely 26 min to execute with very little variance across runs. We present changes/improvements to the RoboSimian system post‐DARPA Robotics Challenge (DRC) (Karumanchi et al., 2016 ). The new architecture has been improved with a focus on scalable autonomy as opposed to semiautonomy as required at the DRC.     

A new software architecture for developing and testing algorithms for space exploration missions

In recent years, planet exploration has received an increasing interest due to the possibility of exploiting planet resources and assuring a human–robotic colonized presence on suitable planetary surfaces. These goals can be reached through the development of smart robots, which are able to work on their own and without requiring a constant human supervision but, at the same time, assuring a great level of safety and reliability. To this aim, the development of effective architectures, concerning both software and hardware issues, could represent a great improvement toward this ambitious objective. This paper presents a novel modular architecture called Test Bench for Robotics and Autonomy (TBRA), the main objective of which is to create a test bench for rover autonomy missions where different implementations of a particular subsystem can be easily tested, while keeping the rest of the system unchanged. Thus, it allows the developers to be able to compare the results of tests and understand which version works better. Such architecture has been built on top of the Workframe, a generic middleware for real-time robotics. This two-layered approach allows the final user to deal only with the TBRA interface, which is designed to be extremely simple to use and takes care of most real-time programming problems, while allowing flexibility in the development, maintenance and future extension of the TBRA itself.     

Fault‐tolerant on‐board computing for robotic space missions

This paper describes an approach to providing software fault tolerance for future deep‐space robotic National Aeronautics and Space Administration missions, which will require a high degree of autonomy supported by an enhanced on‐board computational capability. We focus on introspection‐based adaptive fault tolerance guided by the specific requirements of applications. Introspection supports monitoring of the program execution with the goal of identifying, locating, and analyzing errors. Fault tolerance assertions for the introspection system can be provided by the user, domain‐specific knowledge, or via the results of static or dynamic program analysis. This work is part of an on‐going project at the Jet Propulsion Laboratory in Pasadena, California. Copyright © 2011 John Wiley & Sons, Ltd.     

A survey on the design space of end-user-oriented languages for specifying robotic missions

Software and Systems Modeling - Mobile robots are becoming increasingly important in society. Fulfilling complex missions in different contexts and environments, robots are promising instruments to...     

Autonomy capabilities of European deep space probes

The European Space Agency ESA is currently preparing the two deep space missions, Huygens and Rosetta. This paper reviews the related requirements for autonomous operations in a poorly known environment. While for Huygens emphasis is on the control of the descent through Titan's atmosphere, for Rosetta the safe drilling of material samples in the microgravity environment of a comet is analysed. The control concepts to enable adaptive reactions in an uncertain, remote environment are summarized for the crucial mission phases. The performance results obtained from computer simulations are presented and discussed.     

Towards Model-based Methods for Developing Model-based Systems

Model-based reasoning (MBR) is a means of reasoning about models of all kinds, as appropriate to the task at hand. This includes adaptation of models in response to changes in a problem-solving context or task goals. Thus, MBR exemplifies the characteristics of a smart adaptive system. Constructing appropriate models and matching them to the best inference engines requires a means of describing or defining models. This can be achieved by means of a set of generic model properties. As well as defining models one may decompose the problem domain into a number of tasks to be performed. It then becomes possible to design appropriate models for the domain by mapping these tasks to the model properties. This mapping can be instantiated procedurally, however more generality will result if a model-based approach is taken. As a step towards that goal quality function deployment is investigated as a suitable design method.     

Learning deep kernels in the space of dot product polynomials

Recent literature has shown the merits of having deep representations in the context of neural networks. An emerging challenge in kernel learning is the definition of similar deep representations. In this paper, we propose a general methodology to define a hierarchy of base kernels with increasing expressiveness and combine them via multiple kernel learning (MKL) with the aim to generate overall deeper kernels. As a leading example, this methodology is applied to learning the kernel in the space of Dot-Product Polynomials (DPPs), that is a positive combination of homogeneous polynomial kernels (HPKs). We show theoretical properties about the expressiveness of HPKs that make their combination empirically very effective. This can also be seen as learning the coefficients of the Maclaurin expansion of any definite positive dot product kernel thus making our proposed method generally applicable. We empirically show the merits of our approach comparing the effectiveness of the kernel generated by our method against baseline kernels (including homogeneous and non homogeneous polynomials, RBF, etc...) and against another hierarchical approach on several benchmark datasets.     

Strategies in Model-based Diagnosis

The ability to select suitable diagnostic assumptions and models extends the power of model-based diagnosis for complex systems and can explicitly be modeled by diagnostic strategies. We discuss a framework that allows one to express these strategies as formulas of a meta-language and present a method for designing strategy knowledge bases as well as an efficient straightforward operational semantics for exploiting them.     

Model-based reasoning

In this paper, there will be a particular focus on mental models and their application to inductive reasoning within the realm of instruction. A basic assumption of this study is the observation that the construction of mental models and related reasoning is a slowly developing capability of cognitive systems that emerges effectively with proper contextual and social support. More specifically, we first will identify some key elements of the structure and function of mental models in contrast to schemas. Next, these key elements of modeling will be used to generate some conjectures about the foundations of model-based reasoning. In the next section, we will describe the learning-dependent progression of mental models as a suitable approach for understanding the basics of deductive and inductive reasoning based on models as “tools for thought.” The rationale of mental models as tools for reasoning will be supported by empirical research to be described in a particular section of this paper. Finally, we will turn to the instructional implications of model-based reasoning by discussing appropriate instructional methods to affect the construction of mental models for performing deductive and inductive reasoning.     

Model-based codesign

Hardware-software codesign has been a research topic since the beginning of this decade (1990s), but only now are structured methods emerging that focus on automating design. Unfortunately, to date most codesign approaches leverage performance from individual hardware and software tools, rather than enforcing a structured integration of hardware and software systems simultaneously. A few frameworks have successfully done this integration and have the potential for significant benefits, including reduced time to market, smaller scale design, better likelihood of component reuse, and maximum use of processing power. The article describes a codesign approach that lets developers create models of a formal system representation independently of the hardware and software implementation. The authors' framework, which targets embedded systems, lets developers use simulation based modeling to explore the feasibility of virtual prototypes and then interactively map the specification onto a software-hardware architecture