Application of GA in optimal robot selection for bridge restoration

This paper further expands and tests a method for optimising construction robotʹs kinematic parameters using a genetic algorithm (GA) technique as a tool, proposed earlier by Navon and McCrea [J. Comput. Civ. Eng. 11 (1997) 175]. The approach to selecting an optimal robot is demonstrated on a steel bridge restoration robot. Optimising a robotʹs kinematic parameters is divided into a two-stage problem. Distinctions among the criteria, different ways of assessing the robotʹs behaviour according to these criteria and ranking the criteria according to their importance lie at the base of this division. The robotʹs parameters: (i) major configuration, (ii) minor configuration, (iii) optimal division of the unit length between both links in RRR (revolute configuration) option, (iv) determination of the jointsʹ movement ranges, (v) determination of jointsʹ velocities and accelerations are optimised simultaneously, based on: (i) collision avoidance including arm and wrist singularity avoidance, (ii) percentage of coverage, (iii) dexterity, and (iv) productivity. The procedure for analysis of the inter-relation between the task characteristics and robot design characteristics enables extensions to disparate structures. The GA general function optimisation software is customised here, to be used as the optimisation tool. Comments on the effectiveness of the method and quality of the outcome conclude the paper.