International Review of Automatic Control (I.RE.A.CO.), Vol. 4, n. 6 November 2011 Manuscript received and revised December 2011, accepted November 2011 Fuzzy Logic Control of a Multifingered Hand Robot Using Genetic Algorithm Based on DSP Elham Ataei 1 , Rouhollah Afshari 2 , Mohammad Ali Pourmina 1 Abstract – This paper presents the fuzzy control of a five fingered robot hand by using the genetic algorithm and TMS320F2812 chip. Because of its high-speed performance, its support for multi-motor control and its low power consumption, TMS320F2812 DSP from TI demonstrates itself as an ideal candidate for the five fingered robot hand. There are five fingers and each finger has three degrees of freedom (DOF). All the actuators and electronics are integrated in the finger body and the palm. This robot has five fingers in which all joints are driven by DC motors. At the same time, the multisensory robot hand integrates position, force/torque and temperature sensors. The fuzzy controller dealing with the level of finger control is a multiple-input-multiple-output (MIMO) fuzzy learning controller and is implemented in AVRs microcontrollers. In this paper, the design of the fuzzy controller is base on the genetic algorithm. The whole weight of the hand is about 1.3Kg and the fingertip force can reach 8N. Keywords: Multifinger, Fuzzy Logic, Genetic Algorithm, Hand Robot, DSP Nomenclature DOF = Degrees of freedom FLC = Fuzzy Logic Control GA = Genetic Algorithm DSP = Digital Signal Processor PID = Proportional–Integral–Derivative E= Error r  =Reference Position a  = Actual Position ec = change in errors U = Armature Voltage I. Introduction The five finger robot hands play an ever important role in the service robots and other challenge areas. Recently, various robot hands have been developed so far. Some nice robot hands have been built in the labs and companies, such as the NASA Robonaut Hand [1] the shadow Hand [2] the DLR Hand II [3] and the DLR/HIT Hand I [4]. Generally there are two kinds of hand, one is external actuation hand, where all the actuators are mounted in the forearm (NASA and Shadow), and another internal actuation hand (DLR, HIT), where there needs not any forearm and all the actuators and electronics are integrated in the finger body and the palm. Normally the internal actuation hand body is bigger than the external actuation hand. The external actuation hands are driven by using the tendon cables. The elasticity of the tendon cable causes inaccurate joint angle control, and the long wiring of tendon cables may obstruct the robot motion when the hand is attached to the tip of the robot arm. Moreover, these hands have many problems on the product and the maintenance because it‟s mechanical complexity. To solve these problems, robot hands in which the actuators are built into the hand (e.g., the Belgrade/USC hand by Venkataraman et al. [5], the Omni hand by Rosheim [6], the NTU hand by Lin et al. [7], and the DLR‟s hand by Liu et al. [8]) have been developed. However, these hands have a problem in that their movement is unlike that of the human hand because the number of fingers and the number of joints in the fingers are insufficient. The designed robot is a five-fingered hand driven by built-in DC motors. The control of the multi fingered robot hand is very complicated [9]. The reason is that the degrees of freedom (DOF) of the multi fingered robot hand are too many and the dynamics of the hand are highly nonlinear and coupled. Once the tactile sensors are introduced, it becomes a much more complex control problem that deals with both position and force. Since Zadeh's paper on fuzzy set [10], fuzzy control has become one of the most active and fruitful areas in the applications of fuzzy set theory. Most fuzzy logic control relies on the operators' experiences to design the knowledge bases since the pioneering work of Mamdani [11]. The adaptive fuzzy system equipped with a training algorithm is also applied to the control problem [12]- [14]. Due to the self-organized mechanism, the adaptive fuzzy control can resolve system uncertainty with less heuristic information. In the research of Bekey [l5], a knowledge-based planner was proposed to select gasping postures by reasoning from symbolic information of the target object geometry and the nature of the task. However, the knowledge about grasping in the sense of control was not discussed. Recently, the design of FLC has also been tackled with