Algorithm Development

DRA has extensive algorithm design experience across a wide software landscape. The lead engineers successfully demonstrate the use of the spiral approach in which DRA creates a quantitative description of the problem, models the problem in a meaningful way, evaluates multiple approaches to solving the problem, and then analyzes trade offs between computational constraints and performance requirements. DRA then implements the best solution to the problem and formally quantifies the performance of the solution. Lessons learned from the first development spiral can then be used to improve the models and refine the solution in subsequent spirals. By using this approach, DRA has excelled at developing algorithms to solve problems in the following areas:

  • Image processing and object classification
  • Sub-pixel optical detection
  • Signal detection and characterization
  • Object tracking
  • Control applications
  • Artificial Intelligence and Machine Learning(AI/ML)

Object classification and localization in large Image arrays (6000 x 6000 pixels)

Proficiencies in:

  • Artificial Intelligence (AI)
  • Deep Neural Networks
  • Generative Adversarial Networks (GAN)
  • Support Vector Machines
  • Machine Learning (ML)
  • Long Short-Term Memory (LSTM)
  • Convolutional Neural Networks (CNN)
  • Gaussian Process Regression