|
Current Grants Awarded |
TRACLabs Inc. is integrating the Polyscheme cognitive architecture with an agent-oriented architecture to produce a complete system for providing task-relevant, symbolic data to human stakeholders. Polyscheme is a cognitive architecture for implementing more robust intelligent systems by combining multiple kinds of data structures and algorithms. TRACLabs' agent-oriented architecture is called DCI and was developed via NASA funding to aid distributed human teams in working with autonomous systems. Our integrated system will have the capability to combine a variety of sensory data into a coherent, symbolic world view that is made available to human stakeholders and is defined by an ontology. The system allows for new symbols and categories of symbols to be created with minimal human involvement. The result is a system that easily ties multiple human stakeholders into the cognitive system while at the same time relieving these human stakeholders of routine cognitive tasks. NASA Phase II SBIR/STTR Awards
This SBIR focuses on developing software to assist with mission planning. In Phase II we are targeting mission planning that occurs on the ground prior-to or during mission operations. However, this software could be modified to support on-board planning and re-planning by crew members of their mission activities. The SBIR assumes NASA will have an XML representation for electronic procedures that will reside on-board the spacecraft. It develops extensions to that XML representation that contain planning information, e.g., resources, time requirements, personnel requirements, pre-conditions, etc. Our planning software can be used by crew members or mission operations personnel to create feasible, alternative mission plans given the mission goals, procedures, available resources and available time. Resulting plans can be automatically executed against a simulation to verify and validate it.
This SBIR focuses on developing software that can transform raw telemetry into more abstract, fused information that represents the current or future states of the vehicle. Space vehicles are highly instrumented and generate large amounts of data or telemetry. For this data to be useful to humans monitoring these systems and to automated algorithms controlling these systems, it will need to be converted into more abstract data. This abstracted data will reflect the trends, states and characteristics of the systems and their environments. We are developing a Data Abstraction Architecture (DAA) that allows engineers to design software processes that iteratively converts spacecraft data into higher and higher levels of abstraction. The DAA is a series of mathematical or logical transformations of telemetry data to provide appropriate inputs from a hardware system to a hardware system controller, system engineer, or crew. The DAA also formalizes the relationships between data and control and the relationships between the data themselves. This software has the potential to assist flight controllers, especially those with less experience, in monitoring and controlling space vehicles.
This SBIR provides a software framework to assist ground operators interacting remotely with robots. It aids user understanding of progress on tasks and awareness when important milestones are achieved or problems impede tasks. It also helps users understand current and previous situations by relating operational events to underlying data. Capabilities provided by this framework include (1) computation of robotic health and performance measures, (2) summarization and presentation of these computations, and (3) notification of appropriate personnel when milestones are achieved or performance indicates a problem. This software framework has been evaluated with NASA’s K10 rovers performing reconnaissance and site survey operations during field tests at Haughton Crater in July 2007 and Moses Lake Sand Dunes (WA) in June 2008. It also was evaluated for ground support of a test of remote robotic reconnaissance operations at NASA Ames Research Center in November 2008. This software has the potential to coordinate human and machine activities in space. We also are evaluating this software for monitoring system performance during Exploration analog field tests of lunar robotic operations.
This SBIR is building an innovative manipulation system that includes a modular dexterous manipulator for various mobile platforms and a software control system that seamlessly coordinates motion control of rover and manipulator. The manipulation system has two main components. The primary component is a lightweight, low-power manipulator for mobile platforms. The manipulator will be swiftly reconfigurable with up to seven degrees of freedom (DOF). There will be several different tools available for use at the end effector: some passive, some active. All associated electronics will be internal to the manipulator, requiring only power and data connections externally. Connections between modules will use an innovative "Universal Mating Adapter". The second innovative component is a software control system that coordinates control of the vehicle and manipulator. Such coordination extends the robot's dexterous workspace and facilitates teleoperation by providing the operator with a unified interface.
Robots will play an important role in NASA's exploration activities over the next several decades. They will land on the Lunar surface ahead of humans and help prepare for human exploration. They will explore the Lunar surface, build structures, and move regolith. As humans arrive, these robots will shift to assisting humans in exploration activities. All of these activities require a new generation of robotic vehicles -- capable of flexible, dexterous manipulation -- that can work in closely coordinated teams. Since it will be impossible to tightly script activities on the Lunar surface due to the dynamics of assembly tasks, it is essential that these teams of robots be able to operate in unmodeled environments and unanticipated situations. We expect mobile manipulators to be very important in the development of complex planetary operations and are addressing key issues related to combining mobility, manipulation, and teamwork. Our work focuses on coordinating the use of mobility and manipulator degrees of freedom to achieve a common manipulation purpose. We coordinate multiple mobile manipulators so as to achieve a common goal, such as grasping or manipulating an object so that it can be transported or mated. Procedures are the accepted means of commanding spacecraft. Procedures encode the operational knowledge of a system as derived from system experts, testing, training and experience. NASA has tens of thousands of procedures for Space Shuttle and the International Space Station, which are used daily by both flight controllers and crew. It is expected that the new Constellation vehicles, including Orion, Altair and Lunar habitats, will have thousands of procedures to ensure safe operation. Currently procedures are executed manually using standard command and control displays. We are proposing a new paradigm whereby procedures interact closely with the next generation telemetry and command displays being developed for NASA and with a procedure assistant that can automatically dispatch commands and evaluate telemetry under tight supervision of the operator. The procedure assistant will consist of an interactive procedure display, a procedure assistant executive, a set of procedure support services and an editor for modifying existing procedures or building simple new procedures. In our paradigm procedures will be just like any other component of an integrated suite of mission control tools. This will greatly enhance the efficiency of flight controllers and reduce training costs associated with having a separate set of tools for procedures.
With the advent of the new exploration initiative, the number of customers and missions to be supported by the NASA Space Communications infrastructure will increase dramatically. As well, new antenna types to be developed in support of exploration will increase, thus increasing the complexity of constraints governing the use of space communications assets. In a new concept, the communications architecture will evolve from the present legacy assets but with the addition of new assets. This future architecture will need a radically new user interface paradigm that must allow space communications missions to both unequivocally specify their requests and also iteratively get those requests integrated with those of other users in increasingly crowded bandwidths. It is our contention that such an interface cannot be developed easily with conventional means, but instead is best designed using intelligent agent technologies, resulting in an intelligent space communications scheduling agent for each user/mission. Therefore, to meet the increased scheduling needs we propose to: design and develop software scheduling agents to interface with existing space communications scheduling engines, using local working databases of active schedule possibilities; to incorporate in the agents models of user preferences for communications requests, conflict resolution and notification of schedule changes; to allow the user/mission to vary the autonomy of the scheduling agent; and to imbue the agents with the capability for planful interactions for peer-to-peer resolution of schedule conflicts.
Incidents related to impaired human performance in space operations can be caused by environmental conditions, situational challenges, and operational deficiencies. Detecting, reporting, and correlating related incidents are key to preventing future incidents. NASA has made significant progress in standardizing the reporting of aviation incidents by developing electronic forms for reporting incidents. While such forms improve report consistency, incident data are not represented in a way that enables computer-based reasoning across reports (e.g., automatic linking of related reports.) TRACLabs proposes to develop a human factors incident-reporting tool for gathering incident data, documenting data in incident reports, and archiving incident data. We will define an XML-based semantic language for incident reporting to capture information about human factors incidents, including multi-modal data. We will develop software for authoring incident reports using this language, archiving these reports, and searching the archives using incident semantics. This project is innovative in defining an incident reporting language that uses an ontology-based vocabulary. This enables improved tools for gathering incident data, and for authoring and archiving incident reports. The semantic indexing provided by the use of incident reporting language permits more sophisticated search of archives, including automatic identification of prior incidents potentially relevant to the current incident.
Intelligent robots for planetary exploration produce a wealth of information -- both science data collected by the robots and data about remote robotic operations. The management and analysis of this data provides unique opportunities as well as significant challenges for both science and rover operations, including understanding and summarizing what data have been collected and using this knowledge to improve data access. TRACLabs proposes to develop software for automatically building semantic summaries of data and images collected by remote rovers and using this information to retrieve subsets of this information for manipulation and visualization. We will use these semantic summaries to construct scripts for spatial and event-based data retrieval (e.g., retrieve data collected at a location). This ability to retrieve and manipulate a subset of data relevant to a situation of interest will be used to provide details on demand displays as well as support data exploration starting from a situation or event. Semantic interpretation has focused on document interpretation and database indexing while the proposed approach provides in-line semantic annotation and summarization of data streams. TRACLabs and its partner Carnegie Mellon University bring extensive experience in advanced software development and rover operations enabling integrated software solutions for NASA's planetary exploration.
Satellite intelligence information is being used increasingly for real-time operations. This requires satellites that can be quickly tasked for new objectives and that can respond to opportunistic situations and external threats. The usefulness of satellites and satellite information would be increased if the satellites could respond quickly and effectively with limited ground operator interaction. In addition, because satellites are increasingly important to modern warfare they also face increasing threats from anti-satellite weapons. Thus, they need a means to effectively and autonomously respond to these threats. This proposal offers an integrated planning and scheduling architecture for autonomously managing satellite missions called the Highly Autonomous Mission Manager for Event Response (HAMMER) system. The HAMMER system will allow a satellite to operate and respond to threats even when it is not in communication with the ground or when time constraints require immediate response to threats. The HAMMER system will attempt to meet mission objectives even in the face of threats. HAMMER will prioritize multiple, competing user goals and requests and determine an optimal ordering of satellite tasks to conserve resources and maximize capability. HAMMER will also ensure that the plan is safe from known threats to the satellite.
There are many mundane and repetitive tasks that humans – even children – perform easily, yet are beyond the capabilities of current robotic systems. These tasks involve dexterous manipulation of objects in unstructured environments, where the exact location and/or characteristics of the objects are not known a priori: cleaning and tidying a house, picking fruit, maintaining a lawn and yard, doing dishes, etc. All of these tasks require tight coordination between vision and action. In order for a system to exhibit dexterous hand-eye coordination, it must have four major components. (1) A vision system capable of identifying and localizing target objects or features in real-time; (2) A hand capable of grasping and manipulating the objects; (3) A manipulator arm capable of positioning the hand to the desired 6-degree-of-freedom pose and following desired trajectories; and (4) A control system able to determine in real-time how to use the output of the vision system and other sensors to adjust commands to the arm and hand. TRACLabs is designing each of these systems. |
Building a Coherent World View from Sensory Data
Enhancing NASA's Procedure Representation Language to Support Planning Operations
A Data Abstraction Architecture for Spacecraft Autonomy
A Software Framework for Coordinating Human-Robot Teams
A Field Reconfigurable Manipulator for Rovers
Coordinated Mobile Manipulation for Robotic Material Handling
Embedding Procedure Assistance into Mission Control Tools
Intelligent Agents For Scheduling Space Communications
Semantic Language and Tools for Reporting Human Factors Incidents
Semantic Summarization for Context Aware Manipulation of Data
Autonomous Mission Management for Satellite Systems
A Robot Capable of Highly Dexterous Eye Hand CoordinationHome | Robotics | Automation | Contact | TRACLabs, Inc. © 2009