1 Appropriate Design of Parallel Manipulators J.-P. Merlet and D. Daney INRIA Sophia-Antipolis 2004 Route des Lucioles, 06902 Sophia-Antipolis Cedex, France Email: Jean-Pierre.Merlet@sophia.inria.fr Abstract Although parallel structures have found a niche market in many applications such as machine tools, telescope positioning or food packaging, they are not as successful as expected. The main reason of this relative lack of success is that the study and hardware of parallel structures have clearly not reached the same level of completeness than the one of serial structures. Among the main issues that have to be addressed, the design problem is crucial. Indeed, the performances that can be expected from a parallel robot are heavily dependent upon the choice of the mechanical structure and even more from its dimensioning. In this chapter, we show that classical design methodologies are not appropriate for such closed-loop mechanism and examine what alternatives are possible. 1.1 Introduction Historically, closed-chain structures have attracted the interest of mathematicians as they offer interesting problems. Some theoretical problems linked to this type of structures were mentioned as early as 1645 by Christopher Wren, then in 1813 by Cauchy [1.1] and in 1867 by Lebesgue [1.2]. But clearly at that time, the technology was not able to deal with any practical applications of this type of structures. The very first application was proposed by Gough for a tire test machine [1.3][1.4], although parallel structures were really put in practice in the 1970’s for a flight simulator with the patent of Cappel in 1964 [1.5] and the seminal paper of Stewart [1.6]. Robotics applications were proposed in the early 1980’s [1.7][1.8]. Starting in the 1990’s, parallel kinematic machines (PKM) have started either to be put in use in various domains such as fine positioning devices or to be considered for potential applications such as machine tools. Although parallel structures have found a niche market in many applications, such as machine tools, telescope positioning or food packaging, they are not as successful as expected. In our opinion, the main reason of this relative lack of success is that the study and hardware of parallel structures have clearly not reached the same level of completeness than the one of serial structures. In this chapter, we will address the