Abstract
In order to achieve unmanned construction with a pneumatic caisson method, the present study develops an autonomous operation system using caisson shovels. The shovel mounted on the ceiling slab of the working chamber moves on a rail track and excavates underground based on the trajectory of a skilled human operator. The excavated soil is removed to the aboveground space through the material shaft using earth buckets. Since the rail track constrains the moving range and the shovel cannot always approach the earth bucket, groups of several shovels are organized to perform cooperative work based on a bucket relay system. In the present paper, we proposed a job scheduling algorithm based on terrain recognition, soil mountain selection, excavation trajectory control, and collision avoidance for an autonomous operation. Here, we used a 1/10-scale caisson platform reproducing a practical construction site and performed three cases of cooperative experiments by (a) a group of two shovels, (b) two groups of two shovels, and (c) three groups of two shovels. The results showed that, in case (a), the loading ratio always remained over 100% and the average was 130%. The cooperation of two shovels was achieved. In case (b), sharing an initial soil mountain between two groups, although the loading ratio decreased by 18%, cooperative work between groups as well as that between shovels was achieved without collisions. In case (c), sharing two initial soil mountains among three groups, the loading ratios were 126%, 107%, and 96% for the three groups. The proposed excavation planning showed the feasibility of massive construction using numerous autonomous shovels.
