OFBENet

This repository contains the dataset used in our paper: Orientation-Free Neural Network-Based Bias Estimation for Low-Cost Stationary Accelerometers

Abstract

Low-cost micro-electromechanical accelerometers are widely used in navigation, robotics, and consumer devices for motion sensing and position estimation. However, their performance is often degraded by bias errors. To eliminate deterministic bias terms a calibration procedure is applied under stationary conditions. It requires accelerometer leveling or complex orientation-dependent calibration procedures. To overcome those requirements, in this paper we present a model-free learning-based calibration method that estimates accelerometer bias under stationary conditions, without requiring knowledge of the sensor orientation and without the need to rotate the sensors. The proposed approach provides a fast, practical, and scalable solution suitable for rapid field deployment. Experimental validation on a 13.39-hour dataset collected from six accelerometers shows that the proposed method consistently achieves error levels more than 52% lower than traditional techniques. On a broader scale, this work contributes to the advancement of accurate calibration methods in orientation-free scenarios. As a consequence, it improves the reliability of low-cost inertial sensors in diverse scientific and industrial applications and eliminates the need for leveled calibration.

---

Datasets

Sensors information

Sparkfun website: https://www.sparkfun.com/sparkfun-6-degrees-of-freedom-breakout-lsm6dso-qwiic.html

Memsense webstie: https://memsense.com/products/ms-imu3025/#documentation

Gravity Aligned Dataset

This dataset contains stationary accelerometer recordings collected from four SparkFun ADXL345 IMUs (LSM6DSO), placed in a gravity-aligned orientation. The setup ensures that each sensor’s sensitive axes are aligned with the global reference frame.

Data Collection Setup

• All four IMUs were mounted on a stable, level surface
• Before recording, the sensors’ roll and pitch angles were manually verified to be ~0°, ensuring that gravity acts along a single axis
• The sensors remained completely stationary throughout recording
• All accelerometer readings are expressed in g units
• Total recording time: 8.89 hours

Folder structure:

gravity_aligned_dataset/
├── 1/
│   ├── trial001.csv
│   ├── trial002.csv
│   └── ...
├── 2/
├── 3/
├── 4/

---

Rotated Dataset

The recordings were collected using two SparkFun ADXL345 IMUs (LSM6DSO) together with a high-grade Memsense MS-3025 IMU, which served as the ground-truth reference for rotation angles.

Data Collection Setup

• All three IMUs were mounted on a custom 3D-printed plate designed to keep the sensors rigid and aligned

• The plate was attached to a tripod, allowing controlled adjustments of pitch and roll by tilting the tripod head

• The SparkFun IMUs were logged using a SparkFun Thing Plus ESP32 WROOM running custom firmware

• The Memsense MS-3025 IMU was recorded through its official software suite

Folder structure:

rotated_dataset/
├── sparkfuns/
│   ├── 1.csv
│   ├── 2.csv
│   └── ...
├── memsense/
│   ├── 1.csv
│   ├── 2.csv
│   └── ...

---

If you find the information helpful in your research, please cite our paper:

@misc{levin2025orientationfreeneuralnetworkbasedbias,
      title={Orientation-Free Neural Network-Based Bias Estimation for Low-Cost Stationary Accelerometers}, 
      author={Michal Levin and Itzik Klein},
      year={2025},
      eprint={2511.13071},
      archivePrefix={arXiv},
      primaryClass={cs.RO},
      url={https://arxiv.org/abs/2511.13071}, 
}