RoNIN: Robust Neural Inertial Navigation in the Wild

This video showcase our work on RoNIN: Robust Neural Inertial Navigation in the Wild.
Dataset and code are available now.
Paper: arxiv.org/abs/...
Website: ronin.cs.sfu.ca/
Code: github.com/Sac...

[@michaelsilverhouse2122]

The guy made RONIN text with the trajectory at the end of the video.. Wow!! Impressive :)

[@michaelsilverhouse2122]

would have been better if the guy stops at the end of RONIN's second and last 'N' and waves to camera, but instead he continues to walk :)

[@AssassinGrudge]

Epic, can't wait to read the paper. But please tell me, did you build your work around the name. or it just happen to be like that X')

[@bastianklopfer2652]

Great work! Is it possible to use the network with only IMU data? (raw acceleration and gyr-data, no further data from the phone)

[@hinokikuaimu]

yes, and that's the whole point of the research.

[@omgtruth171]

is the prediction done on mobile device in real time or collecting data first and estimating on desktop?

[@SachiniHerathA]

The evaluations were done offline. But we have developed an app for real-time predictions using RoNIN LSTM version as the model

[@omgtruth171]

@@SachiniHerathA Oh, you mean deploying trained model on mobile device then doing prediction also on board?

[@SachiniHerathA]

@@omgtruth171 yes, the models were trained offline. Prediction can be done online in real time (by deploying the network on mobile device) or offline

[@omgtruth171]

@@SachiniHerathA Thanks!

[@yongshengli3004]

@@SachiniHerathA you made a big progress, could you share this prototype app code for real-time predictions?

[@omgtruth171]

Is there a chance I can download the code?

[@SachiniHerathA]

Hi, we will release the code and data soon