EKF-based self-attitude estimation with DNN learning landscape information

Autor: Ryota Ozaki, Yoji Kuroda
Jazyk: angličtina
Rok vydání: 2021
Předmět:
0209 industrial biotechnology
Control and Optimization
lcsh:T55.4-60.8
Computer science
lcsh:Machine design and drawing
lcsh:Mechanical engineering and machinery
lcsh:Automation
lcsh:Control engineering systems. Automatic machinery (General)
Inference
Angular velocity
02 engineering and technology
Flight simulator
lcsh:Technology
Extended Kalman filter
Attitude estimation
lcsh:TJ212-225
020901 industrial engineering & automation
Artificial Intelligence
lcsh:Technology (General)
0202 electrical engineering
electronic engineering
information engineering
lcsh:Industrial engineering. Management engineering
lcsh:TJ1-1570
lcsh:T59.5
Instrumentation
Ground truth
Artificial neural network
lcsh:T58.5-58.64
business.industry
Covariance matrix
lcsh:T
lcsh:Information technology
Mechanical Engineering
Deep learning
020208 electrical & electronic engineering
Pattern recognition
Mobile robotics
lcsh:TJ227-240
Modeling and Simulation
lcsh:T1-995
Artificial intelligence
business
Zdroj: ROBOMECH Journal, Vol 8, Iss 1, Pp 1-12 (2021)
ISSN: 2197-4225
Popis: This paper presents an EKF-based self-attitude estimation with a DNN (deep neural network) learning landscape information. The method integrates gyroscopic angular velocity and DNN inference in the EKF. The DNN predicts a gravity vector in a camera frame. The input of the network is a camera image, the outputs are a mean vector and a covariance matrix of the gravity. It is trained and validated with a dataset of images and corresponded gravity vectors. The dataset is collected in a flight simulator because we can easily obtain various gravity vectors, although the method is not only for UAVs. Using a simulator breaks the limitation of amount of collecting data with ground truth. The validation shows the network can predict the gravity vector from only a single shot image. It also shows that the covariance matrix expresses the uncertainty of the inference. The covariance matrix is used for integrating the inference in the EKF. Flight data of a drone is also recorded in the simulator, and the EKF-based method is tested with it. It shows the method suppresses accumulative error by integrating the network outputs.
Databáze: OpenAIRE
Externí odkaz: