我们近期发表在《Journal of Computational Design and Engineering》的研究中，提出了一种基于计算设计算法的优化框架，用于四足机器人腿部肢体设计，旨在在全方向行走任务中最大程度地提高能量效率。我们发现，采用过约束设计的Bennett型机器人腿部构型展现出最优的综合能源效率，给四足机器人肢体机构设计带来了新的机遇。
Our recent research published in the Journal of Computational Design and Engineering introduces a computational optimization framework for quadruped robotic leg design, aiming to maximize energy efficiency in various locomotion tasks. We found that the overconstrained Bennett robotic leg configuration shows enhanced energy efficiency, opening new doors for quadruped robotics.
doi: https://doi.org/10.1093/jcde/qwad083
四足机器人一直处于技术进步的前沿，能源效率在其设计中起着至关重要的作用。然而，在机械效率和轨迹规划之间找到完美的平衡一直是一项具有挑战性的任务。在我们的研究中，我们引入了一种全新的计算优化框架，旨在优化四足机器人的基本行走腿部设计，以实现能源效率的最大化。
Legged robots have been at the forefront of technological advancements, with energy efficiency playing a crucial role in their design. However, finding the perfect balance between mechanism efficiency and trajectory planning has long been a challenge. In our study, we introduce a novel computational optimization framework that aims to optimize leg design for basic walking while maximizing energy efficiency in quadruped robots.
计算优化的突破｜A Computational Breakthrough
我们的研究提出了一种用于四足机器人机械臂的计算设计方法，主要关注优化能源高效的步态规划。我们从过度约束的四杆连杆设计出发，泛化了机械臂的运动学。这种方法生成了多样化的设计选择，包括平面、串联、贝内特和球面配置的机械臂，为设计师提供了广泛的选择。
Our research presents a computational design method for quadruped robotic limbs, with a primary focus on optimizing energy-efficient gait planning. We start by generalizing the kinematics of robotic limbs using the overconstrained design of a four-bar linkage. This approach yields a diverse design pool, including robotic limbs with planar, serial, Bennett, and spherical configurations, offering designers a spectrum of options.
最大化能量效率｜Maximizing Energy Efficiency
为了追求能量高效的运动，我们制定了一个计算优化问题，考虑了机器人在前进、侧向和转弯任务中的能源效率。挑战在于确保设计在全方向运动下仍然高效，而我们提出的框架采用了一种无导数方法来实现这一目标。
In our quest for energy-efficient locomotion, we have formulated a computational optimization problem that incorporates the robot’s energy efficiency during forward, lateral, and turning tasks. The challenge was to ensure that the design remains efficient across various locomotion types, and our proposed framework achieved this by using a derivative-free approach.
通过硬件实验验证｜Validation Through Hardware Experiments
为了验证我们的框架，我们使用可重构的四足机器人硬件进行了实验，其中包括可替换的机械臂。结果与我们的预测结果进行了比较，发现原型机器人表现出相似的趋势，具有相对较小的标准化均方误差。对于基本任务的腿部配置优化表明，能源效率可能提高约10%到20%。
To validate our framework, we conducted experiments using reconfigurable quadruped hardware with replacement limbs. The results were compared to our forecasted outcomes, and we found that the prototype robot exhibited a similar trend with a relatively small normalized mean square error. The optimization of leg configuration for primitive tasks showed the potential for improving energy efficiency by approximately 10% to 20%.
释放过约束肢体机构的潜力｜Unlocking the Potential of Overconstrained Limbs
我们研究的一个最重要的发现是过度约束机械臂设计在促进全向运动中提高能源效率的潜力。与通常的做法相反，我们的研究提供了计算设计证据，即贝内特连杆在侧向和转向任务中表现出更高的能源效率，同时在前进行走方面具有竞争力。
One of the most significant findings of our study is the potential advantage of overconstrained robotic limb design for promoting energy efficiency in omni-directional locomotion. Contrary to common practice, our research offers computational design evidence that the Bennett linkages show enhanced energy efficiency during lateral and turning tasks while remaining competitive in forward walking.
本研究的贡献｜Contributions of Our Study
我们的研究提供了一种全面的计算设计框架，用于优化四足机器人的机械臂。在该框架内，我们旨在考虑离散的腿部配置和连杆参数，以帮助设计师生成适用于特定任务要求的可行和优化的机械臂设计。此框架还可以提供有关设计参数（如腿部配置和连杆参数）的宝贵见解。
Our research provides a comprehensive computational design framework for optimizing robotic limbs in quadruped robots. By considering both discrete leg configuration and continuous link parameters within the framework, we aim to assist designers in generating feasible and optimized limb designs for specific task requirements. This framework can also offer valuable insights into the design parameters, such as leg configuration and link parameters.
未来方向｜Future Directions
虽然我们的工作已经显示出了令人鼓舞的结果，但有必要承认其局限性。我们的算法主要关注平坦地形上的全向运动，留下了在更复杂地形和任务中进行进一步探索的空间。未来研究可以引入更多的参数，并探索使用强化学习方法进行协同优化问题。这些进展将为机械臂设计提供更全面的理解，适用于更广泛的机器人应用。
While our work has shown promising results, it is essential to acknowledge its limitations. Our algorithm primarily focuses on omni-directional locomotion on flat terrain, leaving room for further exploration in more complex terrains and tasks. Future research may introduce additional parameters and explore co-optimization problems with reinforcement learning approaches. These advancements will provide a more comprehensive understanding of robotic limb design, catering to a broader spectrum of robotic applications.
总之，我们的研究代表了四足机器人领域的重大进展。我们已经利用计算优化的力量来提高能源效率，从而推动了机械臂的设计和性能。我们的研究结果不仅仅是关于效率，更是关于释放更多多才多艺的机器人系统的潜力，我们的框架为该领域未来的创新铺平了道路。
In conclusion, our research represents a significant step forward in the field of quadruped robotics. We have harnessed the power of computational optimization to enhance energy efficiency, ultimately advancing the design and performance of legged robots. The findings in our study are not just about efficiency; they are about unlocking the potential for more versatile and capable robotic systems, with our framework paving the way for future innovations in the field.
Yuping Gu, Ziqian Wang, Shihao Feng, Haoran Sun, Haibo Lu, Jia Pan*, Fang Wan*, and Chaoyang Song* (2023). “Computational Design Towards Energy Efficient Optimization in Overconstrained Robotic Limbs.” Journal of Computational Design and Engineering, 10: 1-16.
doi: https://doi.org/10.1093/jcde/qwad083
