This thesis "MULTISENSOR INTEGRATION FOR A ROBOT" presents a scheme for combining the various sensor data using artificial intelligence techniques for a robot to perform desired tasks. The sensors used are vision, acoustic and force/torque. Objects are represented by reducing the vision sensor data to a manageable number of symbolic entities. Both 2-D and 3-D object features are considered. A knowledge base is created using frame type of representation which contains the knowledge about the features of the various objects the robot is likely to encounter in its workspace. A matching between the unknown and the known objects is achieved using heuristic search technique. Subgraph isomorphism is checked using bipartite matching to classify partially visible objects. Integrating the various sensor data outputs, commands are generated for the robot. Experimental results demonstrate the validity of the scheme. C programming language is used for the implementation of vision processing algorithms. Object representation and matching algorithms are programmed in LISP.