MIT Designs Dynamic Output Feedback Controller Using LabVIEW and CompactRIO

             The Feedback Control Systems course at MIT focuses on the design and analysis of control systems using classical control and state-space techniques. The course is open to both undergraduate and graduate students and enrolls approximately 20 students each fall. As part of the course, students are required to design and implement roll, pitch, and yaw controllers for physical systems in a series of lab modules. Students design classical controllers using root loci, Bode plots, and other techniques, and develop state-space controllers using linear quadratic regulators (LQR), linear quadratic Gaussian (LQG), and dynamic output feedback (DOFB) designs. Students implement state feedback, state estimation, and dynamic control law design using the LabVIEW Control Design and Simulation Module and the LabVIEW MathScript RT Module. After students validate their controllers through simulation, they deploy their designs using CompactRIO, LabVIEW FPGA, and the LabVIEW Real-Time Module to control a highly nonlinear Quanser 3-DOF helicopter kit.

　　In the fall of the 2010 school year, 42 students were divided into groups of three or four and completed the labs at six different hardware stations. One of the biggest hurdles we experienced in the past semester was getting all the stations set up correctly. The old solution required us to spend a lot of time at the beginning of each semester troubleshooting connections and testing each station. Multiple cables were required to connect the PC to the external data acquisition modules, which complicated the process; the boards connected to the amplifiers amplified the signals from the Quanser kits. With CompactRIO, all sensor and actuator signals can be sent back to the PC over a single Ethernet cable, simplifying the connection and installation steps.

　　The course also makes extensive use of computer-aided control design tools. Students design controllers based on hardware models that ensure the closed-loop system is stable and meets all design requirements. The previous framework, which was built on MathWorks, Inc. Simulink® software, did not provide students with diagnostic tools to test the controllers before deploying them on hardware; most testing was done by students themselves using MathWorks, Inc. MATLAB® software. As a result, a large amount of time in the lab was spent implementing features that did not require hardware, such as diagnostics for controller designs. LabVIEW Control Design and Simulation and the RT Module of LabVIEW MathScript are useful tools for analyzing linear models and assisting students in designing controllers.

　　Throughout the semester, we introduced how to develop inner and outer loop controllers using frequency domain techniques such as Bode and Nyquist plots and state space techniques such as regulators designed via LQR and estimators designed via LQE. Unlike in the past, the LabVIEW front panel provides useful visualizations through 3D graph controls and displays all signal information, making it easy for students to diagnose controllers and update controller designs. The 3D images of the actual kit are very useful, allowing students to compare simulations and real systems side by side to see how they are related. Thanks to this, we effectively demonstrated the difficult concept of model uncertainty and introduced methods to design powerful controllers to compensate for modeling errors.

　　In addition to LabVIEW’s flexibility in adjusting controllers within a complete simulation system, the biggest benefit of using LabVIEW and CompactRIO is the intuitive and easy switching between simulation and reality. Students can simulate and verify their controllers, then immediately deploy them to CompactRIO to control the helicopter by adjusting the front panel controls. Because the simulation structure matches the hardware very closely, simulation can be a good predictor of whether the hardware will succeed, reducing the amount of hardware testing required. This is very effective for managing large class labs, where available lab time is very valuable.

　　In addition to attracting students’ interest, the combination of LabVIEW and CompactRIO has proven to be effective in validating control theory and design methods. Interactive LabVIEW front panels provide an easy way to visualize the system; the ability to probe signals in the block diagram is very useful when debugging the controller. As the semester progressed, students became more and more comfortable modifying the LabVIEW code to meet their needs. As part of the course project, several students designed their own VIs to implement multiple-input, multiple-output (MIMO) controller designs. At the end of the semester, we found that many students used their spare time to participate in our extracurricular competition, which required a helicopter to autonomously cross a virtual obstacle course. At the end of the semester, many students were interested in applying LabVIEW to other MIT projects and actively contacted the course staff.