Locomotion of Linear Actuator Robots Through Kinematic Planning and Nonlinear Optimization

@article{usevitch_locomotion_2020,
  title = {Locomotion of {Linear} {Actuator} {Robots} {Through} {Kinematic} {Planning} and {Nonlinear} {Optimization}},
  volume = {36},
  issn = {1552-3098, 1941-0468},
  url = {https://ieeexplore.ieee.org/document/9106865/},
  abstract = {In this article consider a class of robotic systems composed of high-elongation linear actuators connected at universal joints. We derive the differential kinematics of such robots, and show that any instantaneous velocity of the nodes can be achieved through actuator motions if the graph describing the robot’s configuration is infinitesimally rigid. We formulate physical constraints that constrain the maximum and minimum length of each actuator, the minimum distance between unconnected actuators, the minimum angle between connected actuators, and constraints that ensure the robot avoids singular configurations. We present two planning algorithms that allow a linear actuator robot to locomote. The first algorithm repeatedly solves a nonlinear optimization problem online to move the robot’s center of mass in a desired direction for one time step. This algorithm can be used for an arbitrary linear actuator robot but does not guarantee persistent feasibility. The second method ensures persistent feasibility with a hierarchical coarse-fine planning decomposition, and applies to linear actuator robots with a certain symmetry property. We compare these two planning methods in simulation studies.},
  language = {en},
  number = {5},
  urldate = {2021-03-03},
  journal = {IEEE Transactions on Robotics},
  author = {Usevitch, Nathan S. and Hammond, Zachary M. and Schwager, Mac},
  month = oct,
  year = {2020},
  pages = {1404--1421},
  month_numeric = {10}
}