GTD Unlimited Capabilities

Major projects led by GTD senior personnel include (but are not limited to): multi-target tracking of acoustic objects; structural health monitoring; hyper-spectral target detection and classification; minimizing energy consumption in forge processing; digital radar target detection and classification; and communications simulation. All of these projects required data fusion to facilitate automated “data to decision”.

  • The multi-target acoustic tracking program for unattended ground sensors (UGS) fused GPS, compass, seismic and acoustic data to form measurements. These measurements were then passed into a decision tree that managed an extended Kalman filter. The output of the Kalman filter was then used to make a maximum-likelihood (ML) decision as to whether or not to report a target.
  • GTD has been on the leading edge in developing structural health monitoring systems for more than a decade. This includes developing an optimization strategy for the selection and placement of sensors to satisfy multiple objectives, statistically detecting the presence of damage, and estimating the location and extend of the damage present, all in applications with many non-stationary noise sources. The data from the systems are combined with safety inspection data to form an automated weighted cost/risk decision as to when to perform a full inspection.
  • The hyper-spectral program collected data on stressed crops. The data then underwent feature extraction to reduce the data size. This provided two useful outcomes. First, the reduced data could be more easily analyzed by an operator as it is well known “that large data sets are overwhelming analysts”. Second, the reduced data set allowed the use of a lower-cost sensor for fielded use. The reduced cost sensor could then be used to in a “data to decision” process to automate the maintenance of crops.

  • Optimization of forging processes required the development of mathematical process models and microstructure evolution models along with optimization algorithms to determine precise thermo-mechanical processing profiles and parameter such as die shapes and billet temperatures to achieve final product specifications.

  • The radar programs involved the detection of human and nonhuman targets. In these programs, PCA, wavelet denoising, and optimal deconvolution algorithms (for optimal range resolution and target separation) were used to display target detection and tracks.

  • The communications simulation program required the development of high fidelity communications device models. The effectiveness of the communications link was then analyzed to predict performance.

GTD efforts successfully resulted in: a higher probability of detections and fewer false alarm rates; an increased situational understanding in operational missions and thus support more relevant and informed decisions; and a significantly enhanced ability to navigate and find important relationships and targets in extremely large data sets.

Core Competencies

  1. Data to decision
    1. Probabilistic/Risk based decision-making
      • Decision Trees
      • Maximum Likelihood
      • Bayes
    2. Feature extraction
      • Wavelet
      • PCA
      • Self-Organized Maps
    3. Estimation
      • Parametric models
      • Stochastic signal processing
      • Kalman/Adaptive filters
      • Particle Swarm
    4. Classification
      • Neural networks
      • Hidden Markov Models
      • Discriminant Analysis
      • Bayesian belief
  1. Domain Knowledge
    1. Mechanical Structures
      • Rotorcraft blades
      • Bridges
      • Aircraft
    2. Hyperspectral
      • VNIR
      • SWIR
    3. Atmospheric Acoustics
      • Array design
      • Wind noise
    4. Radar
      • FMCW
      • Pulse
    5. Comms
      • GSM
      • CDMA
      • 802.11
      • DTMF
  1. Real-time Processing
    1. linear
    2. non-linear
      • Levinberg-Marquardt
      • Extended Kalman Filter
    3. custom numerical algorithms
      • c
      • Fortran


GTD Unlimited