Decentralized and Asymmetric Multi-Agent Learning in Construction Sites

TL;DR

A decentralized and asymmetric multi-agent learning approach for construction sites (DAMALCS) and develops two heuristic experts capable of achieving their joint goal optimally, by applying an innovative prioritization method.

Abstract

Multi-agent collaboration involves multiple participants working together in a shared environment to achieve a common goal. These agents share information, divide tasks, and synchronize their actions. Key aspects of multi agent collaboration include coordination, communication, task allocation, cooperation, adaptation, and decentralization. On construction sites, surface grading is the process of leveling sand piles to increase a specific area's height. In this scenario, a bulldozer grades while a dumper allocates sand piles. Our work aims to utilize a multi-agent approach to enable these vehicles to collaborate effectively. To this end, we propose a decentralized and asymmetric multi-agent learning approach for construction sites (DAMALCS). We formulate DAMALCS to reduce expected collisions for operating vehicles. Therefore, we develop two heuristic experts capable of achieving their joint goal optimally by applying an innovative prioritization method. In this approach, the bulldozer's movements take precedence over the dumper's operations, enabling the bulldozer to clear the path for the dumper and ensure continuous operation of both vehicles. Since heuristics alone are insufficient in real-world scenarios, we utilize them to train AI agents, which proves to be highly effective. We simultaneously train the bulldozer and dumper agents to operate within the same environment, aiming to avoid collisions and optimize performance in terms of time efficiency and sand volume handling. Our trained agents and heuristics are evaluated in both simulation and real-world lab experiments, testing them under various conditions, such as visual noise and localization errors. The results demonstrate that our approach significantly reduces collision rates for these vehicles.

Figures (7)

• Figure 1: An example of construction site with five agents: dumper ($agnet_1$), bulldozer ($agnet_2$), compactors ($agnet_3$,$agnet_4$) and an excavator ($agnet_5$)
• Figure 2: DAMALCS - Decentralized and Asymmetric Multi-Agent Learning in Construction Sites. We begin by training each agent individually using BC. Each agent is provided with its own height map and self-position data, from which it generates a predicted trajectory. This self-planned trajectory, along with the positions of other agents, is then fed into our DAMALCS module, that updates and outputs a tuple of collision-free trajectories for both agents, for safe and coordinated movement.
• Figure 3: Visualizing the dumper expert heuristics. The dimensions are for the demonstration only. Loading and Unloading points for the dumper. Fig. \ref{['fig:loading_station_empty']} visualizes the empty dumper at the loading point $L^{Du} = (200,10) [cm]$ described in Eq. \ref{['eq:loading_point']}, and Fig. \ref{['fig:loading_station_full']} visualize a full dumper at the loading point. In fig. \ref{['fig:dumping_area_full']} the dumper is full when arriving to the dumping point $D^{Du}_{1} = (115,145) [cm]$ described in Eq. \ref{['eq:dumping_point']} and in Fig. \ref{['fig:dumping_area_empty']} the dumper is empty after dumping the sand at the dumping point $D^{Du}_{1}$.
• Figure 4: Dozer and Dumper Behaviour Cloning - learning optimal trajectories from dozer and dumper individual heuristic Experts.
• Figure 5: DAMALCS - Decentralized and asymmetric multi-agent learning in construction sites training.