Gradient-Based Trajectory Optimization With Learned Dynamics

被引:4
作者
Sukhija, Bhavya [1 ]
Kohler, Nathanael [1 ]
Zamora, Miguel [1 ]
Zimmermann, Simon [1 ]
Curi, Sebastian [1 ]
Krause, Andreas [1 ]
Coros, Stelian [1 ]
机构
[1] ETH, Dept Comp Sci, Zurich, Switzerland
来源
2023 IEEE INTERNATIONAL CONFERENCE ON ROBOTICS AND AUTOMATION, ICRA | 2023年
关键词
MODEL-PREDICTIVE CONTROL; SYSTEM-IDENTIFICATION; NEURAL-NETWORK;
D O I
10.1109/ICRA48891.2023.10161574
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
Trajectory optimization methods have achieved an exceptional level of performance on real-world robots in recent years. These methods heavily rely on accurate analytical models of the dynamics, yet some aspects of the physical world can only be captured to a limited extent. An alternative approach is to leverage machine learning techniques to learn a differentiable dynamics model of the system from data. In this work, we use trajectory optimization and model learning for performing highly dynamic and complex tasks with robotic systems in absence of accurate analytical models of the dynamics. We show that a neural network can model highly nonlinear behaviors accurately for large time horizons, from data collected in only 25 minutes of interactions on two distinct robots: (i) the Boston Dynamics Spot and an (ii) RC car. Furthermore, we use the gradients of the neural network to perform gradient-based trajectory optimization. In our hardware experiments, we demonstrate that our learned model can represent complex dynamics for both the Spot and Radio-controlled (RC) car, and gives good performance in combination with trajectory optimization methods.
引用
收藏
页码:1011 / 1018
页数:8
相关论文
共 50 条
[21]   A robust gradient-based MPC for integrating real time optimizer (RTO) with control [J].
D'Jorge, Agustina ;
Ferramosca, Antonio ;
Gonzalez, Alejandro H. .
JOURNAL OF PROCESS CONTROL, 2017, 54 :65-80
[22]   A new gradient-based neural network for solving linear and quadratic programming problems [J].
Leung, Y ;
Chen, KZ ;
Jiao, YC ;
Gao, XB ;
Leung, KS .
IEEE TRANSACTIONS ON NEURAL NETWORKS, 2001, 12 (05) :1074-1083
[23]   A Globally Convergent Stochastic Pairwise Conjugate Gradient-Based Algorithm for Adaptive Filtering [J].
Ahmad, Noor Atinah .
IEEE SIGNAL PROCESSING LETTERS, 2008, 15 (914-917) :914-917
[24]   A new gradient-based neural dynamic framework for solving constrained min-max optimization problems with an application in portfolio selection models [J].
Nazemi, Alireza ;
Mortezaee, Marziyeh .
APPLIED INTELLIGENCE, 2019, 49 (02) :396-419
[25]   Trajectory Tracking of Intelligent Vehicle Using Model Predictive Control Based on Neural-dynamics Optimization [J].
Yu, Jimin ;
Huang, Mei ;
Yan, Chuanyou .
2019 CHINESE AUTOMATION CONGRESS (CAC2019), 2019, :588-593
[26]   A hybrid gradient-based and differential evolution algorithm for infinite impulse response adaptive filtering [J].
Yuenyong, Sumeth ;
Nishihara, Akinori .
INTERNATIONAL JOURNAL OF ADAPTIVE CONTROL AND SIGNAL PROCESSING, 2014, 28 (10) :1054-1064
[27]   One-layer gradient-based MPC plus RTO of a propylene/propane splitter [J].
Hinojosa, Aldo Ignacio ;
Ferramosca, Antonio ;
Gonzalez, Alejandro H. ;
Odloak, Darci .
COMPUTERS & CHEMICAL ENGINEERING, 2017, 106 :160-170
[28]   Identifying an inaccessible electrostatic source with gradient-based inverse problem methodology and boundary elements [J].
Arkadan, AA ;
Subramaniam-Sivanesan, S .
IEEE TRANSACTIONS ON MAGNETICS, 1999, 35 (03) :1578-1581
[29]   Gradient-based training and pruning of radial basis function networks with an application in materials physics [J].
Maatta, Jussi ;
Bazaliy, Viacheslav ;
Kimari, Jyri ;
Djurabekova, Flyura ;
Nordlund, Kai ;
Roos, Teemu .
NEURAL NETWORKS, 2021, 133 :123-131
[30]   A Gradient-Based Wind Power Forecasting Attack Method Considering Point and Direction Selection [J].
Jiao, Runhai ;
Han, Zhuoting ;
Liu, Xuan ;
Zhou, Changyu ;
Du, Min .
IEEE TRANSACTIONS ON SMART GRID, 2024, 15 (03) :3178-3192