[X] I've searched other issues and no duplicate issues were found.
[X] I've agreed with the maintainers that I can plan this task.
Description
By default, Autoware relies only on the NDT algorithm as the pose source. However, in environments where NDT performance may be suboptimal, leveraging a GNSS sensor with high accuracy can yield improved results by combining their outputs.
Related Discussion: https://github.com/orgs/autowarefoundation/discussions/4134#discussioncomment-8394375
However, since there isn't a real-time method readily available to detect errors in the NDT algorithm, directly incorporating both GNSS and NDT outputs into an Extended Kalman Filter (EKF) would not be an appropriate approach. There is no suitable method in current autoware to use NDT and GNSS together.
Purpose
It is aimed to develop a suitable method to use GNSS and NDT pose outputs together in localization.
Possible approaches
By pre-processing the pose sources from NDT and GNSS before they enter the EKF, it ensures optimal utilization of incoming poses, leading to more successful outcomes across diverse environments.
If we possess error information for a GNSS sensor, we can dynamically switch between GNSS and NDT based on this knowledge
Definition of done
Developing an efficient method to leverage both NDT and GNSS pose sources effectively
Checklist
Description
By default, Autoware relies only on the NDT algorithm as the pose source. However, in environments where NDT performance may be suboptimal, leveraging a GNSS sensor with high accuracy can yield improved results by combining their outputs. Related Discussion: https://github.com/orgs/autowarefoundation/discussions/4134#discussioncomment-8394375
However, since there isn't a real-time method readily available to detect errors in the NDT algorithm, directly incorporating both GNSS and NDT outputs into an Extended Kalman Filter (EKF) would not be an appropriate approach. There is no suitable method in current autoware to use NDT and GNSS together.
Purpose
It is aimed to develop a suitable method to use GNSS and NDT pose outputs together in localization.
Possible approaches
By pre-processing the pose sources from NDT and GNSS before they enter the EKF, it ensures optimal utilization of incoming poses, leading to more successful outcomes across diverse environments. If we possess error information for a GNSS sensor, we can dynamically switch between GNSS and NDT based on this knowledge
Definition of done
Developing an efficient method to leverage both NDT and GNSS pose sources effectively