Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2025 Jun 18;25(12):3810.
doi: 10.3390/s25123810.

Embedded Sensor Data Fusion and TinyML for Real-Time Remaining Useful Life Estimation of UAV Li Polymer Batteries

Jutarut Chaoraingern  1 , Arjin Numsomran  2
Affiliations

Embedded Sensor Data Fusion and TinyML for Real-Time Remaining Useful Life Estimation of UAV Li Polymer Batteries

Jutarut Chaoraingern et al. Sensors (Basel). .

Abstract

The accurate real-time estimation of the remaining useful life (RUL) of lithium-polymer (LiPo) batteries is a critical enabler for ensuring the safety, reliability, and operational efficiency of unmanned aerial vehicles (UAVs). Nevertheless, achieving such prognostics on resource-constrained embedded platforms remains a considerable technical challenge. This study proposes an end-to-end TinyML-based framework that integrates embedded sensor data fusion with an optimized feedforward neural network (FFNN) model for efficient RUL estimation under strict hardware limitations. The system collects voltage, discharge time, and capacity measurements through a lightweight data fusion pipeline and leverages the Edge Impulse platform with the EON™Compiler for model optimization. The trained model is deployed on a dual-core ARM Cortex-M0+ Raspberry Pi RP2040 microcontroller, communicating wirelessly with a LabVIEW-based visualization system for real-time monitoring. Experimental validation on an 80-gram UAV equipped with a 1100 mAh LiPo battery demonstrates a mean absolute error (MAE) of 3.46 cycles and a root mean squared error (RMSE) of 3.75 cycles. Model testing results show an overall accuracy of 98.82%, with a mean squared error (MSE) of 55.68, a mean absolute error (MAE) of 5.38, and a variance score of 0.99, indicating strong regression precision and robustness. Furthermore, the quantized (int8) version of the model achieves an inference latency of 2 ms, with memory utilization of only 1.2 KB RAM and 11 KB flash, confirming its suitability for real-time deployment on resource-constrained embedded devices. Overall, the proposed framework effectively demonstrates the feasibility of combining embedded sensor data fusion and TinyML to enable accurate, low-latency, and resource-efficient real-time RUL estimation for UAV battery health management.

Keywords: TinyML; UAV Li polymer battery monitoring; real-time embedded systems; remaining useful life (RUL) estimation; sensor data fusion.

PubMed Disclaimer

Conflict of interest statement

The authors declare no conflicts of interest.

Figures

Figure 1
Figure 1
UAV, RUL LiPo battery-embedded AI sensor, and LabView application.
Figure 2
Figure 2
The proposed real-time RUL estimation structure.
Figure 3
Figure 3
Discharge voltages’ (V) degradation trends.
Figure 4
Figure 4
Capacity (mAh) degradation trends.
Figure 5
Figure 5
Remaining useful life (RUL) and other battery-related features.
Figure 6
Figure 6
RUL-related feature histograms: discharge time (s).
Figure 7
Figure 7
RUL-related feature histograms: decrement 3.6–3.4 V (s).
Figure 8
Figure 8
RUL-related feature histograms: capacity (mAh).
Figure 9
Figure 9
Feedforward neural network architecture.
Figure 10
Figure 10
Embedded AI sensor components.
Figure 11
Figure 11
A quad-rotor UAV with RUL LiPo battery-embedded AI sensor.
Figure 12
Figure 12
Flowchart of C/C++ code on embedded AI sensor.
Figure 13
Figure 13
RUL LabView application software.
Figure 14
Figure 14
RUL LabView events trigger state machine diagram.
Figure 15
Figure 15
LabView block diagrams: wait state and timeout event.
Figure 16
Figure 16
LabView block diagrams: wait state and read data event.
Figure 17
Figure 17
LabView block diagrams: wait state and exit event.
Figure 18
Figure 18
LabView block diagrams: display state and read data event.
Figure 19
Figure 19
Feedforward neural network (FFNN).
Figure 20
Figure 20
Training performance and data explorer.
Figure 21
Figure 21
Training and validation loss.
Figure 22
Figure 22
Testing result data explorer.
Figure 23
Figure 23
Scatter plot of estimated RUL testing results.
Figure 24
Figure 24
RUL estimation experiment.
Figure 25
Figure 25
K-nearest neighbors (KNNs).
Figure 26
Figure 26
Random Forest.
Figure 27
Figure 27
RUL estimation performance comparison.
See this image and copyright information in PMC

Similar articles

See all similar articles

References

    1. Sierra G., Orchard M., Goebel K., Kulkarni C. Battery Health Management for Small-Size Rotary-Wing Electric Unmanned Aerial Vehicles: An Efficient Approach for Constrained Computing Platforms. Reliab. Eng. Syst. Saf. 2019;182:166–178. doi: 10.1016/j.ress.2018.04.030. - DOI
    1. Zhang Z., Si X., Hu C., Lei Y. Degradation Data Analysis and Remaining Useful Life Estimation: A Review on Wiener-Process-Based Methods. Eur. J. Oper. Res. 2018;271:775–796. doi: 10.1016/j.ejor.2018.02.033. - DOI
    1. Zhang Y., Yang Y., Li H., Xiu X., Liu W. A Data-Driven Modeling Method for Stochastic Nonlinear Degradation Process with Application to RUL Estimation. IEEE Trans. Syst. Man Cybern. Syst. 2022;52:3847–3858. doi: 10.1109/TSMC.2021.3073052. - DOI
    1. Si X., Zhang Z., Hu C. Data-Driven Remaining Useful Life Prognosis Techniques: Stochastic Models, Methods and Applications. Springer; Cham, Switzerland: 2017. pp. 3–21.
    1. Li L., Saldivar A.A.F., Bai Y., Li Y. Battery Remaining Useful Life Prediction with Inheritance Particle Filtering. Energies. 2019;12:2784. doi: 10.3390/en12142784. - DOI

LinkOut - more resources