Optimization-Inspired Controller Design for Transient Legged Locomotion

Fisher, Callen; Zyl, Joshua van; Govender, Reuben; Patel, Amir

doi:10.1109/icra48506.2021.9560759

Cited by 2 publications

(1 citation statement)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…1) The mass was assumed to be uniformly distributed 2) The pneumatic cylinder was modelled as a bang-bang force with damping [14] and hard-stops to limit the extension and contraction 3) Only planar motion was allowed, with the support rig modelled as an added mass to the body 4) The leg (pneumatic cylinder) and support rig were assumed to attach directly to the CoM of the body Euler-Lagrange methods, in the form of the manipulator equation, were utilized to generate the EoM of the system. These equations were in the form:…”

Section: B Mathematical Modellingmentioning

confidence: 99%

Controlling a low cost bang bang pneumatic monopod

Fisher

Meyer

2022

SAIEE Afr. Res. J.

View full text Add to dashboard Cite

To date there have been great advances in the legged robotics community. However, these platforms are extremely costly to develop and require complex controllers to perform agile motion, limiting their research to well funded institutions, or purely simulation based studies. This research focuses on an extremely low cost robotic monopod platform that consists of a high powered servo motor as well as a pneumatic actuator. Due to the on/off (bang bang) nature of pneumatics, the platform is challenging to mathematically model. Using a reduced order model of the pneumatic actuator, trajectory optimization methods were implemented to generate acceleration, steady-state and deceleration trajectories. These were then analyzed and a simple state machine controller was developed to implement these trajectories on the robotic platform, with comparisons to the simulation results. In order to test the capabilities of the monopod robot, the above method was further extended with the robot running on multiple different surfaces (hard surface as well as two different gravel surfaces). Results are promising and reveal that simple models and controllers are sufficient to generate stable transient motions for a legged robot running on nonuniform terrain.

show abstract

Section: B Mathematical Modellingmentioning

confidence: 99%