This chapter describes the development and implementation of a qualitative geometric representation. The representation facilitates reasoning methods that overcome some of the limitations of AI systems discussed in previous chapters - in particular, the limitations of robots that cannot solve problems using ``commonsense'' spatial reasoning, and the limitations of experimental qualitative reasoning systems that cannot make use of spatial information.
The qualitative geometric representation incorporates ideas and techniques that have been described in the previous two chapters, and are now used in a new overall context. The first part of this chapter discusses issues that need to be considered while developing any spatial representation or geometric modelling system if it is to be useful in robot reasoning. The second part considers how two major solid modelling techniques can be adapted to qualitative methods, and describes how each resulted in a different approach to two dimensional scene description. The third section presents these two different approaches in detail, discussing the advantages and disadvantages of each, and the fourth section discusses further extensions which have not yet been implemented.