Tag Archives: Dynamical Model

Embedding actual knowledge into Deep Learning to improve its reliability

Lutter M, Peters J., Combining physics and deep learning to learn continuous-time dynamics models, The International Journal of Robotics Research. 2023;42(3):83-107 DOI: 10.1177/02783649231169492.

Deep learning has been widely used within learning algorithms for robotics. One disadvantage of deep networks is that these networks are black-box representations. Therefore, the learned approximations ignore the existing knowledge of physics or robotics. Especially for learning dynamics models, these black-box models are not desirable as the underlying principles are well understood and the standard deep networks can learn dynamics that violate these principles. To learn dynamics models with deep networks that guarantee physically plausible dynamics, we introduce physics-inspired deep networks that combine first principles from physics with deep learning. We incorporate Lagrangian mechanics within the model learning such that all approximated models adhere to the laws of physics and conserve energy. Deep Lagrangian Networks (DeLaN) parametrize the system energy using two networks. The parameters are obtained by minimizing the squared residual of the Euler\u2013Lagrange differential equation. Therefore, the resulting model does not require specific knowledge of the individual system, is interpretable, and can be used as a forward, inverse, and energy model. Previously these properties were only obtained when using system identification techniques that require knowledge of the kinematic structure. We apply DeLaN to learning dynamics models and apply these models to control simulated and physical rigid body systems. The results show that the proposed approach obtains dynamics models that can be applied to physical systems for real-time control. Compared to standard deep networks, the physics-inspired models learn better models and capture the underlying structure of the dynamics.

Scheduling of communications between several nodes for better achieving real-time constraints in a distributed control system, and also a very detailed dynamical model of a wheeled vehicle

Naim Bajcinca, Wireless cars: A cyber-physical approach to vehicle dynamics control, Mechatronics, Volume 30, September 2015, Pages 261-274, ISSN 0957-4158, DOI: 10.1016/j.mechatronics.2015.04.016.

A non-conventional drive-by-wireless technology for guidance and control of a redundantly actuated electric car supported by an on-board wireless network of sensors, actuators and control units is proposed in this article. Several optimization-based distributed feedforward control schemes are developed for such powertrain infrastructures. In view of the limitations of the commercial off-the-shelf wireless communication technologies and the harshness of the in-vehicle environments, a pressing design and implementation aspect, in addition to the robustness against information loss, refers to fulfilling the hard real-time computational requirements. In this work, we address such problems by introducing several distributed event-based control schemes in conjunction with adaptive scheduling at the protocol level. Hereby we obtain a simple tuning mechanism to compromise between the outcome accuracy and computation efficiency (i.e., communication traffic intensity). Using simulative evaluations, we demonstrate the viability of the proposed algorithms and illustrate the impact of external interferences in an IEEE 802.15.4 based wireless communication solution.