Learning-Based Speed Estimation from Accelerometer-Only Inertial Sensing
Researchers have proposed a model called CarSpeedNet that can estimate the speed of a vehicle using only data from a smartphone's accelerometer without needing any additional sensors. The model was tested on over 13 hours of real-world driving data and achieved an average error rate of around 0.7 meters per second when given a window of 4 seconds of acceleration data. The study also explored how the length of the input window affects the accuracy of the estimates.
Researchers have proposed a model called CarSpeedNet that can estimate the speed of a vehicle using only data from a smartphone's accelerometer without needing any additional sensors. The model was tested on over 13 hours of real-world driving data and achieved an average error rate of around 0.7 meters per second when given a window of 4 seconds of acceleration data. The study also explored how the length of the input window affects the accuracy of the estimates.
---
Why it matters: This matters to engineers working on autonomous vehicles or mobile sensing applications because it shows that accurate speed estimation can be achieved with minimal hardware requirements, which could lead to more efficient and cost-effective designs.
Source: https://arxiv.org/abs/2401.07468
This article was originally published at: https://arxiv.org/abs/2401.07468