Skip to main content

A Novel Gradient Feature Importance Method for Neural Networks: An Application to Controller Gain Tuning for Mobile Robots

  • 398 Accesses

Part of the Lecture Notes in Electrical Engineering book series (LNEE,volume 793)

Abstract

In the paper, a novel gradient-based feature importance method for neural networks is described. This method is compared to the existing feature importance method using a trained neural network, which predicts the optimal gains in real time, for a steering controller on a mobile robot. The neural network is trained using the Covariance Matrix Adaptation Evolution Strategy (CMA-ES) algorithm, in order to minimize an objective function. From an analysis using the feature importance methods, key inputs are determined, and their contribution to the neural network’s prediction are observed. Furthermore, using a first-order Taylor approximation of the neural network, an improved control law is determined and tested based on the results of the gradient-based feature importance method. This analysis is then applied to an existing neural network using real-world experiments, in order to determine the behavior of the gains with respect to each input, and allows for a glimpse into the neural network’s inner workings in order to improve its explainability.

Keywords

  • Machine learning
  • Neural network
  • Robotics
  • Mobile robot
  • Control theory
  • Gain tuning
  • Adaptive control
  • Explainable artificial intelligence

This is a preview of subscription content, access via your institution.

Buying options

Chapter
USD   29.95
Price excludes VAT (USA)
  • Available as PDF
  • Read on any device
  • Instant download
  • Own it forever
eBook
USD   189.00
Price excludes VAT (USA)
  • Available as EPUB and PDF
  • Read on any device
  • Instant download
  • Own it forever
Softcover Book
USD   249.99
Price excludes VAT (USA)
  • Compact, lightweight edition
  • Dispatched in 3 to 5 business days
  • Free shipping worldwide - see info
Hardcover Book
USD   249.99
Price excludes VAT (USA)
  • Durable hardcover edition
  • Dispatched in 3 to 5 business days
  • Free shipping worldwide - see info

Tax calculation will be finalised at checkout

Purchases are for personal use only

Learn about institutional subscriptions

References

  1. Bakker, E., Nyborg, L., Pacejka, H.B.: Tyre modelling for use in vehicle dynamics studies. Technical report, SAE Technical Paper (1987)

    Google Scholar 

  2. Bojarski, M., et al.: End to end learning for self-driving cars. arXiv preprint arXiv:1604.07316 (2016)

  3. Deremetz, M., Lenain, R., Thuilot, B., Rousseau, V.: Adaptive trajectory control of off-road mobile robots: a multi-model observer approach. In: 2017 IEEE International Conference on Robotics and Automation (ICRA), pp. 4407–4413 (2017). https://doi.org/10.1109/ICRA.2017.7989509

  4. Gunning, D.: Explainable artificial intelligence (XAI). Defense Advanced Research Projects Agency (DARPA), nd Web 2 (2017)

    Google Scholar 

  5. Hansen, N.: The CMA evolution strategy: a tutorial. CoRR abs/1604.00772 (2016). http://arxiv.org/abs/1604.00772

  6. Hill, A., Laneurit, J., Lenain, R., Lucet, E.: Online gain setting method for path tracking using CMA-ES: application to off-road mobile robot control. In: 2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) (2020)

    Google Scholar 

  7. Hill, A., Lucet, E., Lenain, R.: Neuroevolution with CMA-ES for real-time gain tuning of a car-like robot controller. In: Proceedings of the 16th International Conference on Informatics in Control, Automation and Robotics: ICINCO, vol. 1, pp. 311–319. INSTICC, SciTePress (2019). https://doi.org/10.5220/0007927103110319

  8. Hill, A., Lucet, E., Lenain, R.: A new neural network feature importance method: application to mobile robots controllers gain tuning. In: Proceedings of the 17th International Conference on Informatics in Control, Automation and Robotics, ICINCO 2020, pp. 188–194. ScitePress (2020). https://doi.org/10.5220/0009888501880194

  9. Hornik, K., Stinchcombe, M., White, H.: Universal approximation of an unknown mapping and its derivatives using multilayer feedforward networks. Neural Netw. 3(5), 551–560 (1990) https://doi.org/10.1016/0893-6080(90)90005-6. http://www.sciencedirect.com/science/article/pii/0893608090900056

  10. LeCun, Y., Bengio, Y., Hinton, G.: Deep learning. Nature 521(7553), 436–444 (2015)

    CrossRef  Google Scholar 

  11. Lenain, R., Deremetz, M., Braconnier, J.B., Thuilot, B., Rousseau, V.: Robust sideslip angles observer for accurate off-road path tracking control. Adv. Robot. 31(9), 453–467 (2017)

    CrossRef  Google Scholar 

  12. Lenain, R., Thuilot, B., Cariou, C., Martinet, P.: Adaptive and predictive path tracking control for off-road mobile robots. Eur. J. Control. 13(4), 419–439 (2007)

    CrossRef  Google Scholar 

  13. Liaw, A., Wiener, M., et al.: Classification and regression by randomforest. R News 2(3), 18–22 (2002)

    Google Scholar 

  14. Molnar, C.: Interpretable machine learning. Lulu. com (2019)

    Google Scholar 

  15. Mordvintsev, A., Olah, C., Tyka, M.: Deepdream-a code example for visualizing neural networks. Google Res. 2(5) (2015)

    Google Scholar 

  16. Simonyan, K., Vedaldi, A., Zisserman, A.: Deep inside convolutional networks: visualising image classification models and saliency maps (2013)

    Google Scholar 

  17. Suthaharan, S.: Decision tree learning. In: Machine Learning Models and Algorithms for Big Data Classification. ISIS, vol. 36, pp. 237–269. Springer, Boston, MA (2016). https://doi.org/10.1007/978-1-4899-7641-3_10

    CrossRef  MATH  Google Scholar 

  18. Welch, G., Bishop, G.: An introduction to the Kalman filter. Technical report, Chapel Hill, NC, USA (1995)

    Google Scholar 

Download references

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Ashley Hill .

Editor information

Editors and Affiliations

Rights and permissions

Reprints and Permissions

Copyright information

© 2022 The Author(s), under exclusive license to Springer Nature Switzerland AG

About this paper

Check for updates. Verify currency and authenticity via CrossMark

Cite this paper

Hill, A., Lucet, E., Lenain, R. (2022). A Novel Gradient Feature Importance Method for Neural Networks: An Application to Controller Gain Tuning for Mobile Robots. In: Gusikhin, O., Madani, K., Zaytoon, J. (eds) Informatics in Control, Automation and Robotics. ICINCO 2020. Lecture Notes in Electrical Engineering, vol 793. Springer, Cham. https://doi.org/10.1007/978-3-030-92442-3_8

Download citation