Air Learning: a deep reinforcement learning gym for autonomous aerial robot visual navigation

Author:

Krishnan SrivatsanORCID,Boroujerdian Behzad,Fu William,Faust Aleksandra,Reddi Vijay Janapa

Abstract

AbstractWe introduce Air Learning, an open-source simulator, and a gym environment for deep reinforcement learning research on resource-constrained aerial robots. Equipped with domain randomization, Air Learning exposes a UAV agent to a diverse set of challenging scenarios. We seed the toolset with point-to-point obstacle avoidance tasks in three different environments and Deep Q Networks (DQN) and Proximal Policy Optimization (PPO) trainers. Air Learning assesses the policies’ performance under various quality-of-flight (QoF) metrics, such as the energy consumed, endurance, and the average trajectory length, on resource-constrained embedded platforms like a Raspberry Pi. We find that the trajectories on an embedded Ras-Pi are vastly different from those predicted on a high-end desktop system, resulting in up to $$40\%$$ 40 % longer trajectories in one of the environments. To understand the source of such discrepancies, we use Air Learning to artificially degrade high-end desktop performance to mimic what happens on a low-end embedded system. We then propose a mitigation technique that uses the hardware-in-the-loop to determine the latency distribution of running the policy on the target platform (onboard compute on aerial robot). A randomly sampled latency from the latency distribution is then added as an artificial delay within the training loop. Training the policy with artificial delays allows us to minimize the hardware gap (discrepancy in the flight time metric reduced from 37.73% to 0.5%). Thus, Air Learning with hardware-in-the-loop characterizes those differences and exposes how the onboard compute’s choice affects the aerial robot’s performance. We also conduct reliability studies to assess the effect of sensor failures on the learned policies. All put together, Air Learning enables a broad class of deep RL research on UAVs. The source code is available at: https://github.com/harvard-edge/AirLearning.

Funder

Intel Corporation

Publisher

Springer Science and Business Media LLC

Subject

Artificial Intelligence,Software

Reference76 articles.

1. Abadi, M., Agarwal, A., Barham, P., Brevdo, E., Chen, Z., Citro C., Corrado, G. S., Davis, A., Dean, J., Devin, M., Ghemawat, S., Goodfellow, I., Harp, A., Irving, G., Isard, M., Jia, Y., Jozefowicz, R., Kaiser, L., Kudlur, M., Levenberg, J., Mané, D., Monga, R., Moore, S., Murray, D., Olah, C., Schuster, M., Shlens, J., Steiner, B., Sutskever, I., Talwar, K., Tucker, P., Vanhoucke, V., Vasudevan, V., Viégas, F., Vinyals, O., Warden, P., Wattenberg, M., Wicke, M., Yu, Y.,& Zheng, X. (2015). TensorFlow: Large-scale machine learning on heterogeneous systems. Software available from www.tensorflow.org

2. Adiprawita, W., Ahmad, A. S., & Semibiring, J. (2008). Hardware in the loop simulator in UAV rapid development life cycle. CoRR, vol. abs/0804.3874.

3. Ahn, M., Zhu, H., Hartikainen, K., Ponte, H., Gupta, A., Levine, S., & Kumar, V. (2020). Robel: Robotics benchmarks for learning with low-cost robots. In Conference on robot learning (pp. 1300–1313). PMLR.

4. Bakker, B. (2002). Reinforcement learning with long short-term memory. Advances in Neural Information Processing Systems, 5, 1475–1482.

5. Bellemare, M. G., Naddaf, Y., Veness, J., & Bowling, M. (2015). The arcade learning environment: An evaluation platform for general agents. In Proceedings of the 24th international conference on artificial intelligence, IJCAI’15 (pp. 4148–4152). AAAI Press.

Cited by 25 articles. 订阅此论文施引文献 订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献

1. Reinforcement learning-based drone simulators: survey, practice, and challenge;Artificial Intelligence Review;2024-09-05

2. Autonomous UAV navigation using deep learning-based computer vision frameworks: A systematic literature review;Array;2024-09

3. Enhancing Robotic Autonomy and Deep Reinforcement Learning Applications;Advances in Logistics, Operations, and Management Science;2024-06-28

4. NaviSplit: Dynamic Multi-Branch Split DNNs for Efficient Distributed Autonomous Navigation;2024 IEEE 25th International Symposium on a World of Wireless, Mobile and Multimedia Networks (WoWMoM);2024-06-04

5. NaviSlim: Adaptive Context-Aware Navigation and Sensing via Dynamic Slimmable Networks;2024 IEEE/ACM Ninth International Conference on Internet-of-Things Design and Implementation (IoTDI);2024-05-13

同舟云学术

1.学者识别学者识别

2.学术分析学术分析

3.人才评估人才评估

"同舟云学术"是以全球学者为主线,采集、加工和组织学术论文而形成的新型学术文献查询和分析系统,可以对全球学者进行文献检索和人才价值评估。用户可以通过关注某些学科领域的顶尖人物而持续追踪该领域的学科进展和研究前沿。经过近期的数据扩容,当前同舟云学术共收录了国内外主流学术期刊6万余种,收集的期刊论文及会议论文总量共计约1.5亿篇,并以每天添加12000余篇中外论文的速度递增。我们也可以为用户提供个性化、定制化的学者数据。欢迎来电咨询!咨询电话:010-8811{复制后删除}0370

www.globalauthorid.com

TOP

Copyright © 2019-2024 北京同舟云网络信息技术有限公司
京公网安备11010802033243号  京ICP备18003416号-3