**1. Introduction**

Open-pit mining presents benefits such as large-scale production, high resource recovery rates, and minimal environmental impact [1]. The complexity of the working environment within the open-pit mining area and the low efficiency of manual mining necessitate the introduction of autonomous mining systems [2].

In light of the rapid advancement of robotics, big data, artificial intelligence, and 5G technology, we are now observing the emergence of intelligent unmanned dump truck systems [3,4]. Given that around 50% of the gross operating costs in an open-pit mine would be spent on material transport [5], it has been an obvious trend to deploy unmanned transport tools to replace human labor [6]. Figure 1 illustrates that typical unmanned transportation tools in an open-pit mine include unmanned dump trucks, excavators, crushing stations, and charging piles. The excavator is used for mining ore, while the unmanned dump truck is designated for ore transportation. The truck moves to the location of the excavator for ore loading and then delivers it to the crushing station for unloading. Positioned within the mining area, the crushing station primarily serves to crush and pulverize the raw ore to meet the demands of further processing and utilization. When the unmanned dump truck needs recharging, it navigates to the charging pile to undergo the necessary charging process.

**Citation:** Fang, Y.; Peng, X. Micro-Factors-Aware Scheduling of Multiple Autonomous Trucks in Open-Pit Mining via Enhanced Metaheuristics. *Electronics* **2023**, *12*, 3793. https://doi.org/10.3390/ electronics12183793

Academic Editor: Mahmut Reyhanoglu

Received: 12 August 2023 Revised: 4 September 2023 Accepted: 5 September 2023 Published: 7 September 2023

**Copyright:** © 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

**Figure 1.** Schematic on an unmanned transport system in an open-pit mine.

Operating all of the aforementioned devices automatically is difficult because it requires simultaneously considering the features, principles, and capabilities of all devices when generating control commands, otherwise the automated control performance would be worse than that of human laborers [7]. In this sense, dividing the entire control scheme into multiple layers is a practical and feasible solution [8,9]. As shown in Figure 2, a scheduling module first assigns a traverse order for each of the devices; a decision-making module decides how two or more devices interact locally when their nominal trajectories are conflicting [10,11]; a planning module generates a spatio-temporal curve for each device to track [12,13]. This solution is inherently holding a decoupled strategy, i.e., the features, principles, and capabilities of all devices are no longer considered in a simultaneous way. Adopting such a decoupled strategy easily renders the loss of solution optimality, although it reduces the computational burden. Herein, the scheduling module is particularly important because a suboptimal decision made in the scheduling module would largely influence the downstream modules so that there is no chance to achieve optimality in mining operations. This analysis indicates that the scheduling module is important to guarantee the solution quality of an autonomous operating system in an open-pit mine [14]. The goal of this study is to propose a high-precision scheduling methodology with microscopic factors of each device considered, especially temporal factors. Through this, the scheduling method promises to coarsely find an ideal dispatch solution for the downstream modules efficiently without loss of optimality.

**Figure 2.** Overall flowchart of an unmanned transport system.

#### *1.1. Related Work and Motivations*

This subsection reviews the prevalent scheduling methods for dispatching multiple devices (especially transport vehicles) in an open-pit mine or similar scenarios.

In recent years, the scheduling problem in open-pit mines has received considerable attention. Patterson et al. [15] constructed a unique mixed-integer linear programming problem for multi-truck scheduling and used a Tabu Search algorithm to solve it, aiming to minimize the energy consumption of trucks and excavators. Zhang et al. [16], Yuan et al. [17], Wang et al. [18], Bastos et al. [19], Zhang et al. [20], and Bao et al. [21] employed similar strategies to build a model. However, the optimization objectives in these formulated problems only considered fuel consumption while ignoring factors such as consumed time and output amount. Wang et al. [22] proposed a multi-objective optimization (MOO) algorithm for truck scheduling, while Zhang et al. [23] proposed a decomposition-based constrained dominance genetic principle algorithm (DBCDP-NSGA-II) to solve the multiobjective intelligent scheduling problem for trucks in open-pit mines. Ahumada et al. [24], Chang et al. [25], and Afrapoli et al. [26] also built a multi-objective scheduling model. However, a common limitation of refs. [22–26] is that the formulated problem did not consider the refueling or charging requirements of the trucks. Zhang et al. [27] proposed a meta-heuristic search algorithm to solve a mixed-integer programming problem formulated for the concerned multi-truck scheduling scheme and demonstrated by experimentation that this approach improves the energy efficiency of the transport system in open-pit mines. Smith et al. [28] proposed a time-discretized mixed integer programming (MIP) model for the truck scheduling problem in open-pit mines, and a heuristic is used to quickly generate high-quality feasible solutions. However, the proposed model ignored the path planning between loading and unloading spots. Similarly, Zeng et al. [29], de Melo [30], and Yeganejou et al. [31] did not consider the path planning between loading and unloading spots either.

Most previous scheduling models focused on single-objective optimization, particularly energy consumption, and often overlooked the need for multiple optimization objectives. Additionally, these methods focused solely on truck-carrying activities without considering the refueling or charging requirements of the trucks. Path planning between loading and unloading spots, a crucial aspect of scheduling, has also been largely ignored in previous research. As a conclusion of this subsection, the prevalent scheduling methods do not model the concrete dynamics/kinematics and other temporal constraints of the operating devices; thus, they did not account for the actual complexity of operating an open-pit mine.
