仿蚂蚁设计的低成本六足机器人，能自适应复杂地形并识别81类物体。
GiAnt: A Bio-Inspired Hexapod for Adaptive Terrain Navigation and Object Detection
- 模仿蚂蚁运动机制，采用单自由度腿部结构提升越障能力。
- 可攀爬8厘米高障碍物，支持草地、岩石等复杂地形行走。
- 基于Arduino控制，融合机器学习实现81种物体实时识别。
本文介绍了名为GiAnt的仿生六足机器人，其设计灵感源自蚂蚁在多种地形中的高效运动能力。该机器人重1.75公斤，尺寸为310毫米×200毫米×120毫米，采用3D打印与激光切割的轻量化结构。腿部采用单自由度（DOF）连杆曲柄机构，具备出色的越障能力，可轻松跨越高达8厘米的障碍。相比轮式机器人，其腿式结构显著提升了在不平坦和崎岖地面上的适应性。控制基于Arduino实现，通过步态分析优化腿部控制逻辑。集成机器学习与图像处理技术后，可在实时监控系统中识别81种不同物体。GiAnt为科研、探索与测绘任务提供了高适应性且控制简单的低成本解决方案。
原文摘要 · Abstract (English)
This paper presents the design, development and testing of GiAnt, an affordable hexapod which is inspired by the efficient motions of ants. The decision to model GiAnt after ants rather than other insects is rooted in ants' natural adaptability to a variety of terrains. This bio-inspired approach gives it a significant advantage in outdoor applications, offering terrain flexibility along with efficient energy use. It features a lightweight 3D-printed and laser cut structure weighing 1.75 kg with dimensions of 310 mm x 200 mm x 120 mm. Its legs have been designed with a simple Single Degree of Freedom (DOF) using a link and crank mechanism. It is great for conquering challenging terrains such as grass, rocks, and steep surfaces. Unlike traditional robots using four wheels for motion, its legged design gives superior adaptability to uneven and rough surfaces. GiAnt's control system is built on Arduino, allowing manual operation. An effective way of controlling the legs of GiAnt was achieved by gait analysis. It can move up to 8 cm of height easily with its advanced leg positioning system. Furthermore, equipped with machine learning and image processing technology, it can identify 81 different objects in a live monitoring system. It represents a significant step towards creating accessible hexapod robots for research, exploration, and surveying, offering unique advantages in adaptability and control simplicity.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。