Physically consistent state estimation and system identification for contacts

By Svetoslav Kolev, E. Todorov. 2015

52 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Control and Systems Engineering, Mechanical Engineering

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Control and Systems Engineering

In Mechanical Engineering

Citations in the corpus

Closing the Sim-to-Real Loop: Adapting Simulation Randomization with Real World Experience (2018, background): “Overcoming the discrepancy between simulated models and the real world has been addressed through identifying simulation parameters [15], finding common feature rep2019 International Conference on Robotics and Automation (ICRA) Palais des congres de Montreal, Montreal, Canada, May 20-24, 2019”
Learning to Poke by Poking: Experiential Learning of Intuitive Physics (2016, background): “…is to represent the world in terms of a ﬁxed set of physical parameters such as mass, friction coefﬁcient, normal forces etc and use a physics simulator for computing object dynamics (Kolev & Todorov, 2015; Mottaghi et al., 2015; Wu et al., 2015; Hamrick et al., 2011) from this representation.”
Reinforcement learning for non-prehensile manipulation: Transfer from simulation to physical system (2018, unknown): “System identiﬁcation of model parameters was performed in our prior work [5], but modeling errors are difﬁcult to eliminate completely.”
ContactNets: Learning of Discontinuous Contact Dynamics with Smooth, Implicit Representations (2020, background): “However, differentiating through simulation is numerically challenging if discontinuity is approximated with high-curvature [19], and results from these methods have been limited to simulation and quasi-static real-world interaction with highly-compliant objects.”
Residual Policy Learning (2018, methodology): “Learning an accurate dynamics model is another central focus for robotics (termed system identiﬁcation), and has been approached using analytic gradients [23, 24], ﬁnite differences [25] or Bayesian Optimization [7].”
Revisiting active perception (2016, background): “Although some have explored low-level contact models (e.g., Trinkle et al. 2001; Kolev and Todorov 2015, and others), these are too detailed for the level of abstraction where active perception operates.”
Simulation-based design of dynamic controllers for humanoid balancing (2016, background): “However, system identification for these modern physical simulators has not been extensively studied [28].”