Proceedings of 2005 CACS Automatic Control Conference Tainan, Taiwan, Nov. 18-19, 2005 Design of a DSP-based Office Robot Ming-Shyan Wang, Xuan-Kun Lin, Tzu-Chang Shau, and Yung-Chuan Lin Department of Electrical Engineering, Southern Taiwan University of Technology 1, Nan-Tai St. Yung Kang City, Tainan Hsien, Taiwan, 710, E-mail: mswang@mail.stut.edu.tw Abstract This paper aims at the design and implementation of a DSP-based two-wheel robot. We intend the robot to deliver mails and packs point to point along the specified route. Therefore, a fingerprint detection circuit is first considered for delivery security. Secondary, in order to prevent it from coming into collision with people and obstacles on the path, multiple infrared sensors and human body temperature sensors are used along the circumference of the robot. Two brushless dc motors are employed to drive the wheels and sequentially control direction. A DSP-based drive is designed to generate the pulse-width modulation (PWM) signals of the voltage source inverter (VSI). An extension of the frequency zone-based method is employed to decouple the multiple tuning gains in a multi-loop control system so that the proportional gain and integral gain may be adjusted individually. Finally, a video file will demonstrate the scenarios. Key Word: Two-wheel robot, fingerprint detection, frequency zone-based method. 1. Introduction Compared with the defective mechanical commutator of dc motors and lower torque-to-inertia ratio of induction motors, the permanent-magnet brushless motors dominate the servo applications. The permanent-magnet brushless motor holds a sinusoidal back electromotive force (EMF), called permanent-magnet synchronous motor (PMSM), or a trapezoidal EMF, called brushless dc motor (BLDCM). It is widely recognized that the BLDCM is preferable for high-performance servo applications on speed [1]. A DSP-based two-wheel mobile robot is developed to provide service among offices. The function of the designed robot relies on the sensors and/or equipments mounted on it. Figure 1 shows the functional block diagram. The robot is equipped with four wheels, two for driving and two for supplementing. Two DSP-based drives that include DSP chip TMS320F240 steer the wheels and communicate with fingerprint module and infrared circuit. Two geared brushless dc motors with rated power of 40 W and voltage of 24 V play the role of actuating. Four batteries of 12 V provide the power for voltage source inverter. The fingerprint module checks the qualified users to open on-robot mail box and access the mails by recognizing and verifying the fingerprints. The on-robot and in-office infrared circuits transmit and receive message each other to modify the moving direction along the planned path, and human body temperature sensors along the circumference of the robot are used to prevent collision with people and obstacles. 2. Fingerprint Module On Apr. 6, 2000, BioAPI alliance announced the worldwide standard of "minutiae data only" for fingerprint recognition. Fingerprint module, FCP101, from STARTEK Engineering Inc. [2], collocates with software development kits used in the robot. It makes system integrators integrate various fingerprint products fast. STARTEK proprietary software is available in modules that perform image capture, gain control, minutiae extraction, fingerprint enrollment and one-to-one or one-to-many verification matching. The software modules are based on open architecture standards. FCP101 is a DSP module which is embedded with fingerprint verification algorithm in its DSP. It can run all the fingerprint verification functions under the command of a microcontroller, and carries memory which can record data of users' fingerprint templates and in-and-out logs. It's a fantastic solution for those who would like to develop their own fingerprint verification system with special design for the application of access control or time & attendance management in compact size at economical cost level without the support of an extra PC. One DSP drive receives the recognition from the module to finish security process. The robot communicates with the office stations along the planned path to modify its moving direction and trajectory. A station is equipped with an infrared circuit. The circuit has ICs RPM6938 and PT2249 to receive and decode the message from the robot and a microcontroller 89C51 and PT2248 to respond the message through an infrared LED to the robot. Sensors on human-body temperature are used along the circumference of the robot prevent it from coming into collision with people and obstacles on the path. 3. Robot Control A brushless dc motor has a permanent magnet rotor, and the stator windings are wound such that the back is trapezoidal. It therefore requires Proceedings of 2005 CACS Automatic Control Conference Tainan, Taiwan, Nov. 18-19, 2005 rectangular-shaped stator phase currents to produced constant torque, shown in Fig. 2 [3]. The trapezoidal back EMF implies that the mutual inductance between the stator and rotor is nonsinusoidal. The well-known d-q model is not necessarily the best method for modeling. Hence, hall-effect position sensors located every 60 electrical degrees would be suffice. The circuit equation of the motor in phase variables is v a R 0 0 i a v 0 R 0 i b b v c 0 0 R ic La p Lba Lca Lba Lb Lcb Lca i a ea Lcb ib eb Lc ic ec (1) where v j , i j , and e j , j a, b, c , represent the phase j voltage, current, and back EMF, respectively, R is the stator resistance, p d dt is time derivation. Assume that there is no change in rotor reluctances with angle, then La Lb Lc L Lab Lca Lcb M . Applying current balance condition, ia ib ic 0 , equation (1) becomes 0 0 i a e a v a R 0 0 i a L M (2) v b 0 v c 0 R 0 i b 0 R ic p 0 0 LM 0 0 i b e b L M ic ec The electromagnetic torque is (e i eb ib ec ic ) Te a a , r (3) and the motion equation is (T TL Br ) . (4) pr e J The drive system consists of a current loop to adjust the output torque and a velocity loop to manipulate the velocity and direction of the robot. When tuning or designing a multi-loop control system, an extension of the frequency zone-based method decouples the multiple tuning gains so that they may be adjusted individually [4]. The inner loops operate in the next higher frequency zone to the outer loop. After an inner loop is designed, it acts like a low-pass filter within the outer loop. Figure 3 is the speed control block diagram. A simplified model of the current control loop with a bandwidth of 1 Tc is 1 (1 Tc s) . The popular proportional-plus- integral (PI) control is adopted for easier implementation and get zero steady-state error, K Gs ( s ) K P I (5) s On the Bode plot of speed control loop shown in Figure 4, the frequency range is divided into four zones by pi , sc , and 1 / Tc , where sc is the function may be approximated as o Gsc ( s ) Gs ( s ) K K 1 KT P T 1 Tc s J e s Jes (7) where J e is the sum of J M and reflected inertia from load. The proportional gain will be function of sc and found by the following equation J e sc (8) KT and the integral gain is J 2 (9) K I e sc K T It is seen that, if sc is once chosen, K P and K I are adjusted according to the load. KP 4. Results and Conclusions Figures 5 and 6 display the pictures of the robot and drive. The voltages of phases a and b, with trapezoidal shape, of the BLDCM is shown in Fig. 7. Figure 8 presents the speed response measured in TI evaluation system. A video file is shot to demonstrate the scenarios. Generally, mobile robots with a steering wheel (unicycle) or two independent drive wheels are examples with substantial engineering interest. In addition, most wheeled mobile robots can be classified as nonholonomic mechanical systems. Controlling such systems is, however, deceptively simple. The challenge presented by these problems comes from the fact that a motion of a wheeled mobile robot in a plane possesses three degrees of freedom (DOF); while it has to be controlled using only two control inputs under the nonholonomic constraint [5]. This issue is still to be studied. 5. References [1] P. Pillay and R. Krishnan, “Application characteristics of permanent magnet synchronous and brushless dc motors for servo drives,” IEEE Trans. on Ind. Appl., Vol. 37, No. 5, pp. 986-996, 1991. [2] http://www.startek.com.tw/TC/index.htm. [3] P. Pillay and R. Krishnan, “Modeling, simulation, and analysis of permanent-magnet motor drives, Part II: The brushless dc motor drive,” IEEE Trans. on Ind. Appl., Vol. 25, No. 2, pp. 274-279, 1989. designed bandwidth of the speed control loop and (6) pi K I / K P . [4] G. Ellis, Control System Design Guide, Academic Press, San Deigo, California, USA, 2000. Assuming the ratios of sc / pi and 1 /(Tc sc ) are large enough, in the medium frequency range around sc , the magnitude of the open-loop transfer [5] T.-C. Lee, K.-T. Song, C.-H. Lee, and C.-C. Teng, “Tracking control of unicycle-modeled mobile robots using a saturation feedback controller,” IEEE Trans. Contr. Systems Tech., Vol. 9, No. 2, Proceedings of 2005 CACS Automatic Control Conference Tainan, Taiwan, Nov. 18-19, 2005 pp. 305-318, 2001. Acknowledgements The authors would like to express their appreciation to NSC for supporting under contact NSC 93-2213-E- 218-022. 1M Flash Memory RS-232 DSP Fingerprint Recognition Fingerprint Scanner Figure 7 Input voltages of phases a (above) and b RS-232 DSP TMS320F240 Mail Box Circuit Keyboard LCD Infrared Figure 1 Block diagram of robot 0 ia 30 210 150 Figure 8 Speed response 330 ea ib eb ic ec Figure 2 Back EMF and current waveforms of brushless dc motor. 40 dB / dec KP KI / s 20 dB / dec 0 dB pi sc G (s) sc 1/ Tc 20 dB / dec KT Jes Figure 4 Bode diagram of PI control. 1 Tc s 1 40 dB / dec Proceedings of 2005 CACS Automatic Control Conference Tainan, Taiwan, Nov. 18-19, 2005 r* + r Gs (s) 1 1 Tc s KT Figure 3 Speed control block diagram. Figure 5 Picture of robot Figure 6 Picture of drive Te + TL 1 JM s r