无人推土机的作业路径规划方法研究

黎娆

摘 要

随着人工智能技术的显著进步，智能无人系统也应用到了更多领域。在工程领域中，推土机作为最为常见的工程机械之一，也在进行着无人化的升级和转变。为了解决施工中安全、效率的问题，无人机械的路径规划和优化起着至关重要的作用。规划出既符合现场施工要求又能改善施工效率的机械行走路线是完成无人推土机路径规划工作的关键。复杂的场地条件，施工行走规则等因素都会对无人推土机的路径规划产生影响。故本研究目标是基于这些条件约束和规则，有效解决无人推土机在典型施工场景下的作业路径规划问题。

本研究根据推土机在施工中的行走规律，推及铲运和平整作业两种施工场景下无人作业的标准化行走路线，重点分析不同工况下的施工行走规则和路径影响因素，依据这些形成了算法设计中的约束条件和选择策略。在此基础上，有效地建立了土方工程施工中的推土机铲运作业路径规划问题模型和平整作业路径规划问题模型，开发了适用于这两种场景下的无人推土机施工路径规划算法，改进了传统算法的部分性能，使得改进后的算法在铲运作业问题计算时间上减少了一半，也让改进全覆盖路径规划算法在平整作业问题的路径长度和重复率上最大可减少4.8%和8.4%。同时，还通过仿真分析，验证了方法的有效性，分析了不同因素对于算法效果的影响。最后，使用Unity3D进行了无人推土机在这两种场景下的作业路径仿真模拟，并采取一种“游戏”操作的方式可供施工人员参与到路径规划决策中，收集更多推土机师傅操作路径的经验数据。

无人推土机的作业路径规划是工程机械智能化发展的重要一环。本研究提出了基于土方工程机械施工行走规则的无人推土机作业路径规划模型，并设计相应的适用算法为典型场景下的无人推土机作业路径规划提出解决方案。

关键词：无人推土机；土方机械行走规则；施工行走路线；路径规划算法；仿真分析

Abstract

With the development of Artificial Intelligent, unmanned system has been applied to more fields. In the construction field, bulldozer, as one of the most common earthwork machinery, has also been upgraded by unmanned technology. Path optimization of unmanned machinery plays an important role to solve the problems of safety and efficiency in construction. How to plan the path that meets the construction requirements is the key problem. Complex situations and moving rules have a significant impact on the path planning of unmanned bulldozer. This thesis hopes to effectively solve the basic path planning problem of unmanned bulldozer based on these in typical construction scenarios.

The research content is about two construction scenarios: earth-moving and leveling. According to the standardization operation, this thesis focuses on the moving rules and factors under different working conditions, adding constraints and path selection strategies to algorithms design. This study effectively establishes the multi constraint path planning problem model of bulldozer’s earth-moving and leveling, improves the performance of traditional algorithms, and develops two algorithms of unmanned bulldozer path planning suitable for these scenarios, which reduces the earth-moving problem algorithm execution time by half, and the path length and repetition rate in the leveling problem by 4.8% and 8.4%. Through experimental simulation analysis, the effectiveness of the methods are verified, and the effects of different factors on the algorithms are analyzed. Finally, the operation simulation of unmanned bulldozer in these two scenarios is carried out by Unity 3D, and a game-like system is developed, which can collect the empirical path of bulldozer operator, taken as a supplement for on-site investigation in the future.

The path planning of unmanned bulldozer is an important part of the intelligent development of construction machinery. This thesis builds the path planning model of unmanned bulldozer based on the operations and walking rules of earthwork machinery, designs the corresponding applicable algorithms, and puts forward the solutions for the path planning of unmanned bulldozer in typical scenarios.

Key words: unmanned bulldozer, earthmoving machinery walking rules, construction walking routes, path planning algorithm, simulation analysis