Development of new motion control architecture with FPGA-based IC design

Abstract

Most of today’s robots and machine tools are controlled with simple linear control algorithms, such as PID feedback controller. These algorithms are simple to implement but difficult to achieve prestigious tracking performance, especially when the system has highly nonlinear dynamics and in high-speed motions. Tracking errors can be greatly reduced by incorporating dynamics in the control design, which in many cases, however, requires heavy on-line computation and brings difficulty in industry due to limited computation resources when using conventional control architecture. Further, complexity of control algorithms results in a low sampling frequency that degrades the control performance greatly. In this study, a new motion control architecture is proposed to apply many model based dynamic control algorithms more feasibly. Through investigation of existing control algorithms, it is found that many dynamic control algorithms can be partitioned into a linear portion and a nonlinear portion. In the proposed new control architecture, the linear portion with position/velocity feedback represents the major control loop and is realized in the Field Programmable Gate Array (FPGA). The nonlinear portion acts as a dynamic compensation to the linear portion for complex model-related calculations, and is realized in the Digital Signal Processor (DSP). The control gains/parameters in the linear portion in the FPGA are updated by computations in the nonlinear portion in the DSP. Due to the high speed nature of the FPGA, the sampling frequency of the resultant motion control loop can be implemented at rates far greater than that could ever be obtained by a conventional digital controller. In the proposed motion control architecture, the FPGA chip is utilized to achieve more functions than ever before, because the major servo control loop is moved from the DSP to the FPGA. To achieve such goal, a subsequent research on FPGA-based motion control IC is carried out. The proposed motion control IC integrates most of the motion control functions including T-curve trajectory generation. The core of this motion control IC design is a flexible motion control engine, designed with Verilog Hardware Description Language and implemented in an industry standard FPGA provided by Xilinx. The update rates of the current control loop and position/velocity control loop are up to 120 kHz and 40 kHz, respectively. Incorporated with a general-purpose microcontroller or DSP, the proposed motion control IC can be used to provide a simple, compact, low-cost, and effective solution for high-performance motor control. Given that an economical manufacturing cost and high update rate of the control loop can be achieved, it is believed that such motion control ICs will become key components in future motor controller/amplifier. A 3-axis motion control and amplify system is further designed based on the proposed FPGA-based motion control IC technology. Simulation and experiment results show that the proposed control architecture is effective to achieve better motion performance especially for high speed motions. Finally, the proposed motion control system with the FPGA-based IC technology is applied to an industrial robot. The control algorithm used is a simple nonlinear proportional-integral-derivative (N-PID) controller incorporated with a saturated function design. The proposed N-PID controller is formulated by using a new class of saturated function derived from quasi-natural potential function to shape the position and velocity errors. The global asymptotic stability of the controlled system for the set-point position control is proven by Lyapunov’s direct method and LaSalle’s invariance principle. Simulation and experimental results demonstrate the improved performance of the proposed approach over conventional linear PID controller.

Date of Award	15 Feb 2007
Original language	English
Awarding Institution	City University of Hong Kong
Supervisor	Dong SUN (Supervisor)