Autonomous Control Laboratory
To stay up to date with the latest Autonomous Control Laboratory (ACL) news, visit the ACL website. Find an overview of the ACL's research topics below.
We develop efficient robust numerical methods and software to solve convex optimization problems resulting from control applications. This includes development of Interior Point Method (IPM) algorithms and Multi-Parametric Programming (MPP) methods.
Currently we are developing a real-time Primal-Dual IPM algorithms and software for the solution of Second-Order-Cone-Programming (SOCP) problems. Our software is also used to solve the fuel optimal large divert maneuver guidance problem to enable planetary precision landing, which was tested by JPL and NASA Flight Opportunities Program in 2012 and 2013 on a test vehicle, "Xombie", which is built by Masten Aerospace Systems.
TRN (Terrain Relative Navigation)/G-FOLD (Guidance for Fuel Optimal Large Divert) test
light demonstrating onboard autonomous vision based navigation together with real-time op- timization based control, 2014.
Model Predictive Control (MPC)
MPC has been developed as a powerful control method over the last several decades. We use a model of the control system and solve relevant optimal control problems via real-time optimization algorithms. This allows us to maximize performance by explicitly accounting for state and control constraints of the application at hand, unlike classical methods which ensure constraint satisfaction via indirect and sometimes heuristic means, which limit the achievable performance.
Nonlinear/Robust Estimation and Control
Develops the mathematical and computational tools of analysis and design of nonlinear/uncertain feedback control systems. It establishes results to prove stability and robustness properties of feedback control systems with significant sources of nonlinearity and uncertainty.
Distributed Estimation and Control
Establishes the mathematical framework to analyze and synthesize for multi-agent feedback control systems, such as distributed sensor systems, formation flying spacecraft and aircraft, multi-vehicle swarms.
Convexification of Control Problems
Convexification is to express control problems as convex optimization problems, so that their solution becomes tractable, hence can be automated. This allows us to solve complex control problems very efficiently, potentially in real-time. Hence it enables control of autonomous systems and it automates the control design processes allowing us to evaluate a wide range of design options.