Real-Time Control Gets a Boost

Engineers have long struggled to implement complex control systems in real-time. These systems rely on solving mathematical equations quickly and accurately. One type of equation, known as the Riccati equation, is particularly challenging. It's used in many designs, but its high computational demands make it difficult to implement in real-time computing platforms.

The state-dependent Riccati equation scheme, in particular, is criticized for its heavy computational load. This is because it requires solving the Riccati equation at every instant. A team of researchers decided to focus on a specific problem - thrust vector control - to tackle this challenge. They wanted to ensure that their solution was not only efficient but also asymptotically stable.

The team's analysis provided a guarantee for the applicability and stability of their design. This is significant because it avoids the common practice of relying on numerical checks, which can be time-consuming. However, using standard software like MATLAB to check the applicability of the design resulted in a substantial computational burden. This was mainly due to the software's built-in checking routines.

To overcome this limitation, the researchers extended a state-of-the-art Riccati equation solver called the Structure-Preserving Doubling Algorithm. They implemented it on a Field-Programmable Gate Array (FPGA) hardware platform. This move resulted in remarkable computational efficiency and accuracy. The new solver was able to achieve significant reductions in computation time and accuracy residual compared to the MATLAB benchmark.

The researchers also introduced a tuning flexibility feature. This allows practitioners to balance the trade-off between computational speed and accuracy. The tuning flexibility is directly related to the stopping criterion in the Riccati equation-solving process. This process uses the Frobenius norm of the associated Riccati equation's residual matrix. With this new design, engineers can more easily optimize their control systems for real-time applications.