LM Space Systems' Projects:

  • DARPA System F6 Program
    • Phase I: June 2011 
    • In collaboration with: Prasanta Bose (LM-SSC), Gordon Roesler (USC-ISI)

Scientific Systems' Projects:

  • Air Force SBIR "Veracity Evaluation of non-Redundant Information in Flight sYstems (VERIFY)", PIs: Joseph Jackson, Nima Moshtagh
    • Phase I: (Jan 2011 - June 2011)
    • In collaboration with: Prof. Tim McLain (BYU)
  • Abstract: Mixed-criticality UAV systems could benefit from software components which can evaluate the likelihood that non-redundant systems are functioning properly. SSCI will develop a software toolbox called VERIFY which provides integrity management for single-source, non-redundant sub-systems. VERIFY is comprised of both active and passive fault detection methods. It begins with statistical analysis, developing statistical models from data and assuring that observed trends continue. The secondary level fuses data from critical system states and the mission definition to perform contextual analysis using multiple hypothesis testing. At the highest level, system excitation is used for active diagnosis of faulty systems.

  • Navy SBIR "Dynamic Control Surfaces For Extreme Maneuverability of Underwater Vehicles", PI: Nima Moshtagh
    • Phase I: (Dec 2009 - June 2010)
    • In collaboration with: Prof Michael Triantafyllou (MIT), Prof. David Barret (Olin's College), Prof. Andrew Bennett (Olin's College)
  • Abstract: The main goals of the project were (i) developing a formal approach to design and control of biomimetic underwater vehicles; (ii) identifying the challenges in scaling of the current biomimetic designs to larger submarines, and providing appropriate solutions; and (iii) extending the mechanism of closed-loop control in fish and marine mammals to unmanned underwater vehicles. We formulated the control allocation problem for an AUV with flapping foils, and developed a software toolbox to simulate the net effect of flapping foils for increasing maneuverability of a bio-inspired model submarine. We proposed a biomimetic design for a model submarine that respected the scaling laws, and developed a workplan for hardware development and testing.

  •  NASA/JPL SBIR "Topology Control Algorithms & Software For Spacecraft Formation Flying Networks Under Connectivity And Time-delay Constraints", PI: Nima Moshtagh
  • Abstract: The goal of this project was to develop motion planning software for applications to multi-spacecraft missions, which can help NASA engineers design and analyze different formation topologies for future NASA space missions involving formations of large number of spacecraft (TPF-I, The Stellar Imager, etc.). We solved motion planning problems with coupled translational and rotational constraints to optimize fuel and time in the presence of local and global constraints. Our unique approaches include generation of potential functions with global minima, convexification of constraints, and addition of penalty terms for violating constraints. We will deliver a motion planning software for JPL’s formation control testbed (FCT/FAST).

  • NASA/JPL SBIR "Distributed Formation State Estimation Algorithms Under Resource and Multi-Tasking Constraints", PI: Jovan Boskovic
  • Abstract: The estimation accuracy and performance of any embedded algorithm can be significantly lower than expected during execution, because the multi-tasking processor may pre-empt measurement processing and estimation tasks in favor of other tasks. The goal of this project was to develop distributed spacecraft state estimation algorithms that account for real-time multi-tasking processor constraints and delays in the availability of measurements and make the best use of limited available computing resources. We delivered to NASA JPL a novel Anytime Kalman Filter (AKF) architecture that, i) selects and uses the best measurements under given CPU constraints, and ii) continues to improve the accuracy of estimates by opportunistically using any additional CPU resources that become available.

UPenn Projects:

  • ONR Mathematical, Computer and Information Sciences Division, "Vision-based Distributed Coordination of Multi-vehicle Systems", PI: Ali Jadbabaie
    • Publications: RSS 2005, RSS 2008, TRO 2009
    • In collaboration with Prof. Ali Jadbabaie, Prof. Kostas Daniilidis and Dr. Nathan Michael
  • Abstract: We introduced novel ways to generate collective behaviors within a team of mobile robots using only visual sensing. The proposed control laws are distributed, in the sense that only information from neighboring agents is included. Furthermore, the control laws do not rely on the communication or measurement of heading or distance information among neighbors, but instead require measurements of bearing, optical flow and time-to-collision, all of which can be measured using a simple vision sensor. The stability analysis for both fixed and switching topologies are provided. The effectiveness of the control laws are demonstrated on a group of mobile robots.

  • ARO MURI "SWARMS: Scalable sWarms of Autonomous Robots and Mobile Sensors", Penn (leading institution), Yale, UCSB, MIT, Berkeley
    • Publications: CDC 2005, CDC 2007, TAC 2007, CDC 2008
    • In collaboration with: Prof. Ali Jadbabaie
  • Abstract: Different approaches to motion coordination can be categorized as leader-following, virtual structure and consensus approach. In this work, the consensus approach is used to generate motion coordination for robot dynamics evolving on SE(2) (planar formations), SE(3) (flying formations), SO(3) (satellite attitudes) and etc. Design methodologies for constructing distributed control laws are presented that allow one to generate variety of coordinated behaviors such as parallel, circular and line formations for a group of mobile robots. Using Lyapunov techniques the stability of relative equilibriums are studied. The resulting steering laws have simple geometric intuitions which are based on the structure of each particular formation. In all cases, ideas from graph theory and control theory are merged to analyze stability and synthesize the desired control laws.