PhD thesis was developed at the Department of Computer Systems of Latvia University of Agriculture during the period from September 2009 through February 2013. The PhD thesis consists of 194 pages, comprising 8 tables, 39 pictures, 17 formulae, 4 annexes. 247 literature sources were used. The goal of the PhD thesis is to improve the specification development for heterogeneous multi-robot systems during design stage by analyzing the full solution domain instead of testing only a part of possible solutions. In order to achieve the goal of the thesis the list of objectives were defined as follows: 1. perform analysis of specification development methods applied for heterogeneous multi-robot systems; 2. define specification optimization task and its solution concept for heterogeneous multi-robot systems; 3. develop the procedure for finding optimal specification of heterogeneous multi-robot system in full solution domain; 4. develop mission definition technique and its decomposition approach for heterogeneous multi-robot systems; 5. perform the analysis of the size of feasible solution domain of the specification optimization task; 6. implement and experimentally test heuristic search algorithm for initial evaluation of specifications of multi-robot system; 7. analyze possibility to use simulation techniques for fine evaluation of specification of multi-robot system. The content of PhD thesis is structured according to the goal and the tasks of PhD thesis consisting of 6 sections. Section I gives a general overview of multi-robot systems, discusses the state-of-the-art of the multi-robot research domain and indicates specification selection problem for the heterogeneous multi-robot system. Section II defines specification optimization task for multi-robot system, describes the concepts used for optimization and gives an overview of specification optimization procedure developed within the thesis. Section III describes the first and the second steps of the specification optimization procedure, defines mission decomposition approach into components and tasks. Section IV refers to the third and the fourth steps of the specification optimization procedure, provides analysis of the domain of feasible solutions, defines formulas for estimating the size of the domain, and introduces CaMBot-Gen software used for the analysis. Section V refers to the fifths step of the procedure and defines initial evaluation of solution candidates using heuristic methods, introduces GAMBot-Eva software implementation of genetic algorithm based heuristic search, describes development of genetic representation of the solution domain, and defines the model for the estimation of total costs of ownership.