IPNL-POLYU / UrbanNavDataset

UrbanNav:An Open-sourced Multisensory Dataset for Benchmarking Positioning Algorithms Designed for Urban Areas
385 stars 62 forks source link

about GT #68

Closed nothing371442 closed 4 days ago

nothing371442 commented 6 days ago

Hello, thank you very much for your work! May I ask why Novatel SPAN-CPT can still output centimeter level accuracy in high difficulty scenarios such as urban canyons? Have you done any post-processing work on Novatel SPAN-CPT output to optimize the ground truth? In my understanding, Novatel SPAN-CPT has high-precision inertial navigation and high-precision RTK, but high-precision RTK is not suitable for challenging scenarios. Can high-precision inertial navigation (another set of sensors besides IMU sensors) be used alone to obtain high-precision, long-term, and error free positioning results? thank you!

DarrenWong commented 4 days ago

Thank you for your interest in our dataset. We collect GNSS and IMU raw measurements from the SPAN-CPT and postprocess the data using state-of-the-art Inertial Explorer software from NovAtel, which maximizes the accuracy of the trajectory by processing forward and backward in time, performing a backward smoothing step, and combining the results.

Regarding your question "Can high-precision inertial navigation (another set of sensors besides IMU sensors) be used alone to obtain high-precision, long-term, and error free positioning results?"

Ans: We observed that our existing SPAN-CPT might not be good enough to provide long-term drift-free performance in super-dense urban or tunnel environments. However, it should be good enough to provide the GT in existing datasets. We also working on combining the measurements of SPAN-CPT and LiDAR/camera to achieve long-term and drift-free positioning results in urban areas. Please stay tuned!

For more details related to GT, you can refer to Section 3.1 of the Ubrannav paper

nothing371442 commented 4 days ago

Very clear, thank you very much!