NeOR: neural exploration with feature-based visual odometry and tracking-failure-reduction policy

Abstract

Embodied visual exploration is critical for building intelligent visual agents. This paper presents the neural exploration with feature-based visual odometry and tracking-failure-reduction policy (NeOR), a framework for embodied visual exploration that possesses the efficient exploration capabilities of deep reinforcement learning (DRL)-based exploration policies and leverages feature-based visual odometry (VO) for more accurate mapping and positioning results. An improved local policy is also proposed to reduce tracking failures of feature-based VO in weakly textured scenes through a refined multi-discrete action space, keyframe fusion, and an auxiliary task. The experimental results demonstrate that NeOR has better mapping and positioning accuracy compared to other entirely learning-based exploration frameworks and improves the robustness of feature-based VO by significantly reducing tracking failures in weakly textured scenes.

References

[1] YamauchiB. A frontier-based approach for autonomous exploration[C]. Proceedings 1997 IEEE International Symposium on Computational Intelligence in Robotics and Automation CIRA’97. Towards New Computational Principles for Robotics and Automation, July 10–11, 1997, Monterey, USA, 1997, New York, IEEE: 146-151

[2] CHAPLOT D S, GANDHI D, GUPTA S, et al. Learning to explore using active neural SLAM[EB/OL]. (2020-4-10) [2024-3-11]. https://arxiv.org/pdf/2004.05155.

[3] RamakrishnanS K, Al-HalahZ, GraumanK. Occupancy anticipation for efficient exploration and navigation[C]. 16th European Conference on Computer Vision, August 23–28, 2020, Glasgow, UK, 2020, New York, Springer International Publishing: 400-418

[4] BigazziR, LandiF, CascianelliS, et al.. Focus on impact: indoor exploration with intrinsic motivation[J]. IEEE robotics and automation letters, 2022, 7(2): 2985-2992

[5] GeorgakisG, BucherB, ArapinA, et al.. Uncertainty-driven planner for exploration and navigation[C]. 2022 International Conference on Robotics and Automation (ICRA), May 23–27, 2022, Philadelphia, USA, 2022, New York, IEEE: 11295-11302

[6] YangX, YuC, GaoJ, et al.. Save: spatial-attention visual exploration[C]. 2022 IEEE International Conference on Image Processing (ICIP), October 16–19, 2022, Bordeaux, France, 2022, New York, IEEE: 1356-1360

[7] LIU S, SUGANUMA M, OKATANI T. Symmetry-aware neural architecture for embodied visual navigation[J]. International journal of computer vision, 2023: 1–17.

[8] PartseyR, WijmansE, YokoyamaN, et al.. Is mapping necessary for realistic pointgoal navigation?[C]. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, June 18–24, 2022, New Orleans, USA, 2022, New York, IEEE: 17232-17241

[9] CaoY, ZhangX, LuoF, et al.. Unsupervised visual odometry and action integration for pointgoal navigation in indoor environment[J]. IEEE transactions on circuits and systems for video technology, 2023, 33(10): 6173-6184

[10] Mur-ArtalR, TardósJ D. ORB-SLAM2: an open-source SLAM system for monocular, stereo, and RGB-D cameras[J]. IEEE transactions on robotics, 2017, 33(5): 1255-1262

[11] ZhangJ, LiuJ, ChenK, et al.. Map recovery and fusion for collaborative augment reality of multiple mobile devices[J]. IEEE transactions on industrial informatics, 2020, 17(3): 2081-2089

[12] JinS, MengQ, DaiX, et al.. Safe-nav: learning to prevent pointgoal navigation failure in unknown environments[J]. Complex & intelligent systems, 2022, 8(3): 2273-2290

[13] NaveedK, AnjumM L, HussainW, et al.. Deep introspective SLAM: deep reinforcement learning based approach to avoid tracking failure in visual SLAM[J]. Autonomous robots, 2022, 46(6): 705-724

[14] DaiX Y, MengQ H, JinS, et al.. Camera view planning based on generative adversarial imitation learning in indoor active exploration[J]. Applied soft computing, 2022, 129: 109621

[15] BartolomeiL, TeixeiraL, ChliM. Semantic-aware active perception for UAVs using deep reinforcement learning[C]. 2021 IEEE/RSJ International Conference on Intelligent Robots and Systems, September 28–30, 2021, Prague, Czech Republic, 2021, New York, IEEE: 3101-3108

[16] VaswaniA, ShazeerN, ParmarN, et al.. Attention is all you need[C]. Advances in Neural Information Processing Systems, December 4–9, 2017, Long Beach, USA, 2017, New York, Curran Associates Inc.: 6000-6010

[17] RadfordA, KimJ W, HallacyC, et al.. Learning transferable visual models from natural language supervision[C]. International Conference on Machine Learning, July 18–24, 2021, Virtual, 2021, New York, PMLR: 8748-8763

[18] PlacedJ A, StraderJ, CarrilloH, et al.. A survey on active simultaneous localization and mapping: state of the art and new frontiers[J]. IEEE transactions on robotics, 2023, 39(3): 1686-1705

[19] BonettoE, GoldschmidP, PabstM, et al.. Irotate: active visual SLAM for omnidirectional robots[J]. Robotics and autonomous systems, 2022, 154: 104102

[20] ZhangH, WangS, LiuY, et al.. EFP: efficient frontier-based autonomous UAV exploration strategy for unknown environments[J]. IEEE robotics and automation letters, 2024, 9(3): 2941-2948

[21] HuangJ, ZhouB, FanZ, et al.. FAEL: fast autonomous exploration for large-scale environments with a mobile robot[J]. IEEE robotics and automation letters, 2023, 8(3): 1667-1674

[22] YuwenX, ZhangH, YanF, et al.. Gaze control for active visual SLAM via panoramic cost map[J]. IEEE transactions on intelligent vehicles, 2022, 8(2): 1813-1825

[23] YangY, ZhangJ, QianW, et al.. Autonomous exploration for mobile robot in three dimensional multi-layer space[C]. International Conference on Intelligent Robotics and Applications, July 5–7, 2023, Hangzhou, China, 2023, Singapore, Springer Nature Singapore: 254-266

[24] DengX, ZhangZ, SintovA, et al.. Feature-constrained active visual SLAM for mobile robot navigation[C]. 2018 IEEE International Conference on Robotics and Automation (ICRA), May 21–25, 2018, Brisbane, Australia, 2018, New York, IEEE: 7233-7238

[25] HeK, ZhangX, RenS, et al.. Deep residual learning for image recognition[C]. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, June 26–July 1, 2016, Las Vegas, USA, 2016, New York, IEEE: 770-778

[26] CHO K, VAN MERRIËNBOER B, GULCEHRE C, et al. Learning phrase representations using RNN encoder-decoder for statistical machine translation[EB/OL]. (2014-9-3) [2024-3-11]. https://arxiv.org/pdf/1406.1078.

[27] YeJ, BatraD, DasA, et al.. Auxiliary tasks and exploration enable objectgoal navigation[C]. Proceedings of the IEEE/CVF International Conference on Computer Vision, October 10–17, 2021, Montreal, Canada, 2021, New York, IEEE: 16117-16126

[28] SCHULMAN J, WOLSKI F, DHARIWAL P, et al. Proximal policy optimization algorithms[EB/OL]. (2017-8-28) [2024-3-11]. https://arxiv.org/pdf/1707.06347.

[29] SavvaM, KadianA, MaksymetsO, et al.. Habitat: a platform for embodied AI research[C]. Proceedings of the IEEE/CVF International Conference on Computer Vision, October 27-November 2, 2019, Seoul, Korea, 2019, New York, IEEE: 9339-9347

[30] XiaF, ZamirA R, HeZ, et al.. Gibson ENV: real-world perception for embodied agents[C]. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, June 19–21, 2018, Salt Lake City, USA, 2018, New York, IEEE: 9068-9079

RIGHTS & PERMISSIONS

Tianjin University of Technology