目次

Sensor fusion

Point estimation of particle filter of multimodal posteriori probability distribution

It is now possible to estimate the state using multiple sensors due to miniaturization and high performance of sensors. In addition, multiple sensors may be installed to improve redundancy. However, installing multiple sensors may increase the failure rate.
Nishida laboratory studies the integration of data from multiple sensors using particle filters against these problems. In the particle filter, you can get the probability distribution of states, but in practice you need to extract the only estimate from the probability distribution. A general point estimation method can not give an appropriate estimate when it becomes a multimodal probability distribution. Therefore, we are doing research to extract an appropriate estimated value from multimodal probability distribution.
Nishida_Lab