Design and Simulation of a Multi-Robot Architecture for Large-scale Construction Projects

[ Watch Movie ]

A flexible system architecture, MARSala, that allows fora team of mobile robots to construct motion support structures in large and unstructured environments using local interactions. The proposed architecture provides a means for formulating a construction project as a motion support structure construction problem.

Navigable Path Planning

The augmented local map is used by the

Navigable Path Planner

to compose a plan to build a navigable path from the local origin to the local goal.

The augmented local map consists of the local heightmap, navigability map, local goal, local origin and MARS gap map. The cells with green, yellow, blue and red background colors depict navigable, occupiable, reachable and un-occupiable positions. The cells with green and red foreground colors (and underlined) depict the local start and local goal positions.

The

Local Map Synthesis

provides consistency for the local construction planner in the Global deposition bounds, Target location and Return path. This allows for a decentralized coordination between robots as the actions taken locally are consistent across robots since they depend only on the target location and the local environment.

Scenario A: Collectively building a navigable path

In this scenario, a large, mobile payload system with conservative motion parameters is to be transported to an inaccessible target location in a large unstructured environment. A team of agile builder robots capable of depositing material in their environment are deployed to build a navigable path to the target location for the mobile payload. In the image shown here, each builder robot deposits material of a different color and the path is highlighted in yellow.

The robots were deployed on a topographic model of a portion of Valles Marineris on the planet Mars. We ran the tests for robot teams of different sizes and measured the number of simulation time steps required to reach the target location.

Scenario B: Terrain levelling

A team of robots is tasked to level an uneven terrain as part of a site preparation project.

Nonetheless, MARSala allows each robot to build in reaction to their local environment. Hence, they are inherently coordinated by the environment state, maintaining progress but may not guaranteeing completion.