State estimation for tensegrity robots

Caluwaerts, Ken; Bruce, Jonathan; Friesen, Jeffrey M.; SunSpiral, Vytas

doi:10.1109/icra.2016.7487331

Cited by 30 publications

(16 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For analyzing the systematic error reported by [12], [13], [14], [15], range measurements with line-of-sight conditions and no immediate antenna surroundings were taken using different antenna types in an effort to qualitatively asses their influence on the error. Two Partron Dielectric Chip antennas on DWM1000 modules [16] were placed 1.7 m apart.…”

Section: B Influence Of Antenna Designmentioning

confidence: 99%

“…Motivated by the systematic range error reported by [12], [13], [14], [15], who all used off-the-shelf UWB modules, this paper identifies the range error estimation as one of the keys towards more accurate low-cost, small-size, off-theshelf localization systems. The main contributions are the analysis of the ranging error, the suggested UWB measurement model, and an experimental evaluation.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Ultra-wideband range measurement model with Gaussian processes

Ledergerber

D’Andrea

2017

2017 IEEE Conference on Control Technology and Applications (CCTA)

View full text Add to dashboard Cite

Abstract-This paper presents an ultra-wideband range measurement model based on Gaussian processes. An analysis of the range measurement error with off-the-shelf ultra-wideband radio modules reveals a strong correlation between the reported error and the relative pose of the two ranging modules. A Gaussian process is trained for capturing this correlation and is included in the measurement model. Its effectiveness and real-time applicability are experimentally demonstrated on a quadrocopter platform.

show abstract

Section: B Influence Of Antenna Designmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Ultra-wideband range measurement model with Gaussian processes

Ledergerber

D’Andrea

2017

2017 IEEE Conference on Control Technology and Applications (CCTA)

View full text Add to dashboard Cite

show abstract

“…The dynamics derivation contains the translations and rotations that map the rigid body states onto point-mass positions. Since an analytical dynamics model is needed for the model-predictive controllers, prior work in the field including kinematics-only models [20], [36], [21] and numerical methods [35], [15], [18] could not be used.…”

Section: State-space Modelmentioning

confidence: 99%

“…This rigid body formulation (eqn [32][33][34][35][36][37][38][39][40][41][42][43]. can be combined with the minimum force density optimization problem (eqn [27][28][29].…”

mentioning

confidence: 99%

Model-Predictive Control With Inverse Statics Optimization for Tensegrity Spine Robots

Sabelhaus

Zhao

Zhu

et al. 2021

IEEE Trans. Contr. Syst. Technol.

View full text Add to dashboard Cite

Robots with flexible spines based on tensegrity structures have potential advantages over traditional designs with rigid torsos. However, these robots can be difficult to control due to their high-dimensional nonlinear dynamics. To overcome these issues, this work presents two controllers for tensegrity spine robots, using model-predictive control (MPC), and demonstrates the first closed-loop control of such structures. The first of the two controllers is formulated using only state tracking with smoothing constraints. The second controller, newly introduced in this work, tracks both state and input reference trajectories without smoothing. The reference input trajectory is calculated using a rigid-body reformulation of the inverse kinematics of tensegrity structures, and introduces the first feasible solutions to the problem for certain tensegrity topologies. This second controller significantly reduces the number of parameters involved in designing the control system, making the task much easier. The controllers are simulated with 2D and 3D models of a particular tensegrity spine, designed for use as the backbone of a quadruped robot. These simulations illustrate the different benefits of the higher performance of the smoothing controller versus the lower tuning complexity of the more general input-tracking formulation. Both controllers show noise insensitivity and low tracking error, and can be used for different control goals. The reference input tracking controller is also simulated against an additional model of a similar robot, thereby demonstrating its generality.

show abstract

“…In our case, the full state x t can only be obtained through either simulation or the use of an external state estimator system on the physical SUPERball robot [38]. In contrast, we choose an observation o t that can be calculated directly from the sensors on the robot itself.…”

Section: A Additional Experimental Detailsmentioning

confidence: 99%

Deep reinforcement learning for tensegrity robot locomotion

Zhang

Geng

Bruce

et al. 2017

2017 IEEE International Conference on Robotics and Automation (ICRA)

Self Cite

View full text Add to dashboard Cite

Tensegrity robots, composed of rigid rods connected by elastic cables, have a number of unique properties that make them appealing for use as planetary exploration rovers. However, control of tensegrity robots remains a difficult problem due to their unusual structures and complex dynamics. In this work, we show how locomotion gaits can be learned automatically using a novel extension of mirror descent guided policy search (MDGPS) applied to periodic locomotion movements, and we demonstrate the effectiveness of our approach on tensegrity robot locomotion. We evaluate our method with realworld and simulated experiments on the SUPERball tensegrity robot, showing that the learned policies generalize to changes in system parameters, unreliable sensor measurements, and variation in environmental conditions, including varied terrains and a range of different gravities. Our experiments demonstrate that our method not only learns fast, power-efficient feedback policies for rolling gaits, but that these policies can succeed with only the limited onboard sensing provided by SUPERball's accelerometers. We compare the learned feedback policies to learned open-loop policies and hand-engineered controllers, and demonstrate that the learned policy enables the first continuous, reliable locomotion gait for the real SUPERball robot. Our code and other supplementary materials are available from http://rll.berkeley.edu/drl_tensegrity * These authors contributed equally to this work.

show abstract

State estimation for tensegrity robots

Cited by 30 publications

References 10 publications

Ultra-wideband range measurement model with Gaussian processes

Ultra-wideband range measurement model with Gaussian processes

Model-Predictive Control With Inverse Statics Optimization for Tensegrity Spine Robots

Deep reinforcement learning for tensegrity robot locomotion

Contact Info

Product

Resources

About