The Engineering Meetings Board has approved this paper for publication. It has successfully completed SAE's peer review process under the supervision of the session organizer. This process requires a minimum of three (3) reviews by industry experts. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior written permission of SAE.

Flexibility at the joint of a manipulator is an intrinsic property. Even “rigid-joint” robots, in fact, possess a certain amount of flexibility. Previous experiments confirmed that joint flexibility should be explicitly included in the model when designing a high-performance controller for a manipulator because the flexibility, if not dealt with, can excite system natural frequencies and cause severe damage. However, control design for a flexible-joint robot manipulator is still an open problem. Besides being described by a complicated system model for which the passivity property does not hold, the manipulator is also underactuated, that is, the control input does not drive the link directly, but through the flexible dynamics. Our work offers another possible solution to this open problem. We use three-layer neural networks to represent the system model. Their weights are adapted in real time and from scratch, which means we do not need the mathematical model of the robot in our control algorithm. All uncertainties are handled by variable-structure control. Backstepping structure allows input efforts to be applied to each subsystem where they are needed. Control laws to adjust all adjustable parameters are devised using Lyapunov’s second method to ensure that error trajectories are globally uniformly ultimately bounded. We present two state-feedback schemes: first, when neural networks are used to represent the unknown plant, and second, when neural networks are used to represent the unknown parts of the control laws. In the former case, we also design an observer to enable us to design a control law using only output signals—the link positions. We use simulations to compare our algorithms with some other well-known techniques. We use experiments to demonstrate the practicality of our algorithms.

Accurate air-fuel ratio control is required for good engine performance and low emission in diesel dual fuel engine. Two actuators directly affect the ratio are the air throttle and the EGR valve. Maximum air throttle opening is favorable to minimize pumping loss, and the EGR valve opening should follow closely the values in a well-tuned map to minimize emission. In the past, the two actuators were either controlled separately or simultaneously to achieve the air-fuel ratio set point without much consideration on the actuators' opening positions. We proposed a logic that alternated between actuating the air throttle and the EGR valve to maintain optimum air throttle opening. Since each actuator was controlled one at a time, the overall control system was simplified, yet any advanced controller could be applied to increase the accuracy of each actuator. Experiments on four-cylindrical diesel-dual-fuel engines showed that the air throttle opening was optimized at all time, whereas the EGR valve opening followed closely the values in a well-tuned map. The air-fuel ratio was also accurately regulated with widest range possible. Both new-European-driving-cycle and set point changes tests were performed on engine and chassis dynamometers, which demonstrated the effectiveness of the proposed method.

Input shaping technique has been applied to flexible-joint robot to suppress its residual vibration from fast point-to-point movement. Input shaping performance deteriorates when the knowledge of the mode parameters of the robot is not accurate. Several robust input shapers were proposed at the expense of longer move time. A novel input shaping system, consisting of a quantitative feedback controller, a feed-forward reference model, and a simple zero-vibration (ZV) input shaper, is proposed in this paper. Advantages over the existing robust input shapers include toleration of substantially larger amount of uncertainty in the mode parameters, shorter move time that does not increase with insensitivity, application to nonlinear and time-varying systems, and suppression of vibration induced by disturbance and noise.

We present a state-feedback control of a two-link flexible-joint robot. First, we obtain desired control laws from Lyapunov's second method. Then, we use three-layer neural networks to learn unknown parts of the desired control laws. In this way, the control algorithm does not require the mathematical model representing the robot. We use smooth variable structure controller to handle the uncertainties from neural network approximation and external disturbances. To show the effectiveness and practicality of this control algorithm, we performed an experiment on one of the robots in our laboratory.