rpng / MINS

An efficient and robust multisensor-aided inertial navigation system with online calibration that is capable of fusing IMU, camera, LiDAR, GPS/GNSS, and wheel sensors. Use cases: VINS/VIO, GPS-INS, LINS/LIO, multi-sensor fusion for localization and mapping (SLAM). This repository also provides multi-sensor simulation and data.
GNU General Public License v3.0
469 stars 78 forks source link

IMU and wheel odometer of this work #45

Open supersupercoconut opened 1 week ago

supersupercoconut commented 1 week ago

Thanks for your valuable work! I have some questions regarding this SLAM algorithm:

  1. Could you please clarify whether this work utilizes a 6-axis or 9-axis IMU sensor?
  2. How the wheel odometry impacts the Z-axis accuracy of the system? Does it enhance or diminish the vertical accuracy of the pose estimation?
WoosikLee2510 commented 1 week ago

Hi, @supersupercoconut .

  1. When you say 6 axis is it mean {gyro, accel}, and 9 axis means {gyro, accel, mag}? If that's the case, MINS uses a 6-axis. If you want to include {mag}, I suggest using {mag} separately as an orientation update method.
  2. MINS supports multiple assumptions for wheel odometry (see: https://github.com/rpng/MINS/blob/ed6cf522845e09c8ee33da58f27aad46aeba6337/mins/src/update/wheel/WheelTypes.h#L27-L32). One common mode is Wheel3DAng which allows you to leverage planar motion constraint which helps reducing z-axis drift.
supersupercoconut commented 1 week ago

Hi, @supersupercoconut .

  1. When you say 6 axis is it mean {gyro, accel}, and 9 axis means {gyro, accel, mag}? If that's the case, MINS uses a 6-axis. If you want to include {mag}, I suggest using {mag} separately as an orientation update method.
  2. MINS supports multiple assumptions for wheel odometry (see: https://github.com/rpng/MINS/blob/ed6cf522845e09c8ee33da58f27aad46aeba6337/mins/src/update/wheel/WheelTypes.h#L27-L32 ). One common mode is Wheel3DAng which allows you to leverage planar motion constraint which helps reducing z-axis drift.

Yes, when I refer to the 6-axis IMU, I'm talking about the gyroscope and accelerometer. I would like to confirm if the data from the /joint_state topic is derived from the fusion of the chassis IMU and a wheel odometer?

WoosikLee2510 commented 1 day ago

In my experience, there are different cases. For example, the KAIST dataset provides direct wheel angular velocity, which you can multiply by wheel radius to get each wheel's velocity. On the other hand, Husky was providing odometry fused with IMU, as you said.