Real-Time-Position Prediction Algorithm for Under-Actuated Robot Manipulator Using of Artificial Neural Network

Abstract

Robot manipulators, in general, are required to have the same number of actuators as the number of joints to obtain full control. In the case of under-actuated robots, this condition is not satisfied which make the behavior of that class of robots very difficult to be predicted. Under-actuated robots can be a better design choice for robots in space and other industrial applications, their advantages over fully actuated robots led to many studies to predict their behavior (Yu et al., 1998; Berkemeier & Fearing, 1999; Spong, 1995; Ono et al., 2001; Nakanishi et al., 2000; Funda et al., 1996; Luca et al., 2000; Luca & Oriolo, 2002; Arai & Tachi, 1991; Mukherjee & Chen, 1993;Yu et al., 1993;Bergerman et al., 1995; Mahindrakar et al., 2006; Muscato, 2006; Begovich et al., 2002). As a first advantage, a light-weight and low power consumption manipulator can be made. This feature is required in low cost automation and space robots. Second, they can easily overcome actuator failure due to unexpected accident. The under-actuated manipulator could be the model of the direct drive manipulator that has some failed joints; such fault-tolerant behavior is highly desirable for robots in remote or hazardous environments (Yu et al., 1998). Other interesting applications include the Acrobot (Berkemeier & Fearing, 1999; Spong, 1995), the gymnast robots (Ono et al., 2001), the brachiating robots (Nakanishi et al., 2000), and surgical robots (Funda et al., 1996). The mathematical complexity and wide variety of applications have kept under-actuated robots an area of open research. (Luca et al., 2000; Luca & Oriolo, 2002) have investigated the behavior of a 2R manipulator moving in a horizontal plane with a single actuator at the first joint, neglecting joint friction which is not easy to achieve in real world as it involves high manufacturing cost. Trying to overcome that problem, some researchers have implemented additional equipments such as breaks at the passive joint (Arai & Tachi, 1991; Mukherjee & Chen, 1993; Yu et al., 1993; Bergerman et al., 1995). In this case, the brake can generate torque that means after all that kind of systems is considered some kind of actuator. So, it will be difficult to consider that robot as an under-actuated manipulator. Motivated by this problem, (Yu et al., 1998) have investigated the dynamic characteristics of a two-link manipulator in view of global motion including joint friction by proposing a mathematical model; they have found that the manipulator can be positioned if the friction