Optimizing DNN Architectures for High Speed Autonomous Navigation in GPS Denied Environments on Edge Devices

Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 11671)


We address the challenge of high speed autonomous navigation of micro aerial vehicles (MAVs) using DNNs in GPS-denied environments with limited computational resources; specifically, we use the ODROID XU4 and the Raspberry Pi 3. The high computation costs of using DNNs for inference, particularly in the absence of powerful GPUs, necessitates negotiating a tradeoff between accuracy and inference. We address this tradeoff by employing sparsified neural networks. To obtain such architectures, we propose a novel algorithm to find sparse “sub networks” of existing pre trained models. Contrary to existing pruning-only strategies, our proposal includes a novel exploration step that efficiently searches for a different, but identically sparse, architecture with better generalization abilities. We derive learning theoretic bounds that reinforce our empirical findings that the optimized network achieves comparable generalization to the original network. We show that using our algorithm it is possible to discover models which, on average, have upto 19x fewer parameters than those obtained using existing state of the art pruning methods on autonomous navigation datasets, and achieve upto 6x improvements on inference time compared to existing state of the art shallow models on the ODROID XU4 and Raspberry Pi 3. Last, we demonstrate that our sparsified models can complete autonomous navigation missions with speeds upto 4 m/s using the ODROID XU4, which existing state of the art methods fail to do.

Supplementary material

488940_1_En_36_MOESM1_ESM.pdf (4.2 mb)
Supplementary material 1 (pdf 4261 KB)


  1. Boroujerdian, B., Genc, H., Krishnan, S., Faust, A., Reddi, V.J.: Why compute matters for UAV energy efficiency? (2018)Google Scholar
  2. Delmerico, J., Scaramuzza, D.: A benchmark comparison of monocular visual-inertial odometry algorithms for flying robots. Memory 10, 20 (2018)Google Scholar
  3. Frankle, J., Carbin, M.: The lottery ticket hypothesis: training pruned neural networks. arXiv preprint arXiv:1803.03635 (2018)
  4. Genc, H., Zu, Y., Chin, T.W., Halpern, M., Reddi, V.J.: Flying IoT: toward low-power vision in the sky. IEEE Micro 37(6), 40–51 (2017)CrossRefGoogle Scholar
  5. Giusti, A., et al.: A machine learning approach to visual perception of forest trails for mobile robots. IEEE Robot. Autom. Lett. 1(2), 661–667 (2016)CrossRefGoogle Scholar
  6. Iandola, F.N., Han, S., Moskewicz, M.W., Ashraf, K., Dally, W.J., Keutzer, K.: Squeezenet: Alexnet-level accuracy with 50x fewer parameters and< 0.5 mb model size. arXiv preprint arXiv:1602.07360 (2016)
  7. LeCun, Y., Denker, J.S., Solla, S.A.: Optimal brain damage. In: Advances in neural information processing systems. pp. 598–605 (1990)Google Scholar
  8. Li, H., Kadav, A., Durdanovic, I., Samet, H., Graf, H.P.: Pruning filters for efficient convnets. arXiv preprint arXiv:1608.08710 (2016)
  9. Liu, Z., Sun, M., Zhou, T., Huang, G., Darrell, T.: Rethinking the value of network pruning. arXiv preprint arXiv:1810.05270 (2018)
  10. Loianno, G., Scaramuzza, D., Kumar, V.: Special issue on high-speed vision-based autonomous navigation of uavs. Journal of Field Robotics 35(1), 3–4 (2018)CrossRefGoogle Scholar
  11. Loquercio, A., Maqueda, A.I., del Blanco, C.R., Scaramuzza, D.: Dronet: Learning to fly by driving. IEEE Robotics and Automation Letters 3(2), 1088–1095 (2018)CrossRefGoogle Scholar
  12. Mohta, K., Sun, K., Liu, S., Watterson, M., Pfrommer, B., Svacha, J., Mulgaonkar, Y., Taylor, C.J., Kumar, V.: Experiments in fast, autonomous, gps-denied quadrotor flight. In: 2018 IEEE International Conference on Robotics and Automation (ICRA). pp. 7832–7839. IEEE (2018)Google Scholar
  13. Molchanov, P., Tyree, S., Karras, T., Aila, T., Kautz, J.: Pruning convolutional neural networks for resource efficient transfer learning. arXiv preprint arXiv:1611.06440 (2016)
  14. Park, J., Boyd, S.: A semidefinite programming method for integer convex quadratic minimization. Optimization Letters 12(3), 499–518 (2018)MathSciNetCrossRefGoogle Scholar
  15. Quigley, M., Mohta, K., Shivakumar, S.S., Watterson, M., Mulgaonkar, Y., Arguedas, M., Sun, K., Liu, S., Pfrommer, B., Kumar, V., et al.: The open vision computer: An integrated sensing and compute system for mobile robots. arXiv preprint arXiv:1809.07674 (2018)
  16. Zhang, D., Wang, H., Figueiredo, M., Balzano, L.: Learning to share: Simultaneous parameter tying and sparsification in deep learning (2018)Google Scholar

Copyright information

© Springer Nature Switzerland AG 2019

Authors and Affiliations

  1. 1.Robert Bosch Centre for Cyber-Physical SystemsIndian Institute of ScienceBengaluruIndia
  2. 2.Department of Computer Science and EngineeringIndian Institute of TechnologyHyderabadIndia
  3. 3.Department of Computer Science and AutomationIndian Institute of ScienceBengaluruIndia

Personalised recommendations