Planar Orientation Control and Torque Maximization Using a Swarm With Global Inputs
Planar Orientation Control and Torque Maximization Using a Swarm With Global Inputs
复制标题
DOI：
10.1109/tase.2019.2925908
复制
发表时间：
2019-07
影响因子：
5.6
通讯作者：
Shiva Shahrokhi;Lillian Lin;Aaron T. Becker
中科院分区：
文献类型：
--
作者：
Shiva Shahrokhi;Lillian Lin;Aaron T. Becker
本文研究了大量颗粒对长径比杆的扭矩。所有粒子都被全局信号推向同一个方向。我们计算的力和扭矩产生的三个典型的位置分布的一群：均匀，三角形，和正常的。该模型表明，对于一个枢轴杆的均匀分布产生的最大扭矩小群体的标准偏差，但正态分布最大扭矩大的标准偏差。在仿真中，我们使用这些结果来设计比例微分控制器，以定向刚性对象。最后，我们展示了多达97个硬件机器人的实验，以评估我们的理论在实践中。从业者注意-工作空间的杂乱会阻止大的被操纵粒子操纵物体和机动，而小的粒子可以通过这种杂乱。一个小粒子产生的力比一个大粒子小，所以要产生同样的力，需要更多的粒子。它们的小尺寸使得机载传感和计算变得困难。因此，它们通常由共享控制输入控制。操纵由共享控制输入驱动的粒子群的对象是一项具有挑战性的任务，但当需要设置对象的最终方向时，它甚至更具挑战性。包括组装和交付在内的许多应用都需要物体的特定方向。只有一个转向粒子的扭矩控制很容易：通过在尽可能远离枢轴点的位置推动对象来最大化扭矩。然而，一群粒子在物体上的不同位置产生力。本文研究了如何使用一群粒子形状在三个典型的位置分布的扭矩最大化。假设每个接触物体的粒子传递相等的力，工作受到限制，但硬件实验验证了在施加扭矩时考虑群体分布的必要性。在未来的工作中，我们将研究如何随机粒子之间的接触的影响力和扭矩的传输，并检查在3-D空间中的控制。
This paper studies the torque applied by a large number of particles on a long aspect-ratio rod. The particles are all pushed in the same direction by a global signal. We calculate the force and torque generated by three canonical position distributions of a swarm: uniform, triangular, and normal. The model shows that for a pivoted rod the uniform distribution produces the maximum torque for small swarm standard deviations, but the normal distribution maximizes torque for large standard deviations. In the simulation, we use these results to design proportional-derivative controllers to orient rigid objects. We conclude showing the experiments with up to 97 hardware robots to evaluate our theory in practice. Note to Practitioners—Workspace clutter can prevent large steered particles from being able to manipulate objects and maneuver, while smaller particles can pass through this clutter. A small particle produces less force than a big particle, so to produce the same force, more are needed. Their small size makes onboard sensing and computation hard. Therefore, they are often controlled by a shared control input. Manipulating objects with a swarm of particles actuated by a shared control input is a challenging task, but it is even more challenging when the object’s final orientation needs to be set. Many applications including assembly and delivery require a specific orientation of the object. Torque control with only one steered particle is easy: maximize torque by pushing on the object at a location as far from the pivot point as possible. However, a swarm of particles contributes force at different places on the object. This paper studies how to maximize torque using a swarm of particles shaped in three canonical position distributions. The work is limited by assuming each particle touching the object transmits equal force, but hardware experiments validate the necessity to consider swarm distribution when applying torque. In the future work, we will investigate how stochastic contacts between particles effects force and torque transmission, and examine control in 3-D space.