Biomedical Signal Processing and Control 8 (2013) 382–390 Contents lists available at SciVerse ScienceDirect Biomedical Signal Processing and Control j o ur nal hom epa ge: www.elsevier.com/locate/bspc Technical note Hybrid control strategies for a five-finger robotic hand Cheng-Hung Chen ∗ , D. Subbaram Naidu Measurement and Control Engineering Research Center, School of Engineering, Idaho State University, 921 South 8th Avenue, Stop 8060, Pocatello, ID 83209, USA a r t i c l e i n f o Article history: Received 18 August 2012 Received in revised form 8 February 2013 Accepted 11 February 2013 Available online 16 March 2013 Keywords: Robotic hand Hybrid control Fuzzy logic PD control Adaptive neuro-fuzzy inference system Soft computing Prosthetic hand a b s t r a c t This paper presents a hybrid controller of soft control techniques, adaptive neuro-fuzzy inference system (ANFIS) and fuzzy logic (FL), and hard control technique, proportional-derivative (PD), for a five-finger robotic hand with 14-degrees-of-freedom (DoF). The ANFIS is used for inverse kinematics of three-link fingers and FL is used for tuning the PD parameters with 2 input layers (error and error rate) using 7 triangular membership functions and 49 fuzzy logic rules. Simulation results with the hybrid of FL-tuned PD controller exhibit superior performance compared to PD, PID and FL controllers alone. © 2013 Elsevier Ltd. All rights reserved. 1. Introduction Hard computing/control (HC) techniques can be used at lower- level control for accuracy, precision, stability and robustness and comprise proportional-derivative (PD) control [1], proportional- integral-derivative (PID) control [2,3], optimal control [3–6], adaptive control [7–10], etc. with specific applications to robotic hand devices. Recently, the authors conducted an overview of control strategies for robotic and prosthetic hands [11,12]. How- ever, our previous works [1–3,13] for a robotic hand showed that PID controller resulted in undesirable feature of overshooting and oscillation, which were also demonstrated by Subudhi and Mor- ris [14] in a two-link flexible robot manipulator and Liu et al. [15] in a 6-degrees-of-freedom (DoF) underwater robot (autonomous underwater vehicle). Unlike HC, soft computing (SC) or computational intelligence (CI) techniques are meant to adapt to an environment under impre- cision, uncertainty, partial truth and approximation [16]. Recently, the review paper of L. Magdalena has analyzed, compared and dis- cussed some definitions of SC found in literature [17]. Unlike the lower-level control of HC, SC can be used at high-level control of the overall mission where human involvement and decision making is of primary importance. SC is an emerging field based on synergy and seamless integration of neural networks (NN), fuzzy logic (FL) and ∗ Corresponding author. Tel.: +1 617 388 1509. E-mail addresses: chenchen@isu.edu, nthumary@gmail.com (C.-H. Chen). optimization methods, such as genetic algorithms (GA) and par- ticle swarm optimization (PSO) [1,16–23]. Several SC techniques have been applied to robotic and prosthetic hands. Arslan et al. [24] developed the biomechanical model with a tendon configu- ration of the 3-DoF index finger of the human hand and the fuzzy sliding mode controller in which a FL unit tuned the slope of the sliding surface to generate the required tendon forces during clos- ing and opening motions. Kato et al. [25] expressed the reaction of brains to the adaptable prosthetic system for a 13-DoF electromyo- graphic (EMG) signal controlled prosthetic hand with an EMG pattern recognition learning by artificial NN. Kamikawa and Maeno [26] used GA to optimize locations of pivots and grasping force and designed one ultrasonic motor to move 15 compliant joints for an underactuated five-finger prosthetic hand. Khushaba et al. [27] developed a PSO-based method for myoelectrically controlled prosthetic devices. However, the artificial hands had limitation on precision grasping, such as grasping a screw or needle. To overcome the limitation, the accuracy and effectiveness of fingertip trajectory and control systems need to be optimized. The brain analogy corresponds to the fusion of HC and SC tech- niques. We therefore propose hybrid intelligent control strategies with the integrated structure by blending [19,20] the upper-level control of SC techniques and lower-level control of conventional HC techniques. Fig. 1 shows the proposed integration or fusion of SC and HC methodologies can solve problems that cannot be solved satisfactorily by using either HC or SC methodology alone and can lead to high performance, robust, autonomous and cost- effective missions, such as accuracy and effectiveness of fingertip 1746-8094/$ – see front matter © 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.bspc.2013.02.003