基于膨胀视焦点的生物启发碰撞检测神经网络

doi:10.19678/j.issn.1000-3428.0252184

摘要/Abstract

摘要： 利用传统计算机视觉技术来解决复杂场景中的碰撞检测问题是一项十分艰巨的任务，尤其是面临虚碰撞干扰时模型误检率较高而准确率较低。针对该问题，论文基于哺乳动物视网膜分层结构特性，借助灵长类动物大脑皮层中央前回多感觉皮质区（PZ）神经元对特定视野区的危险感知特性，提出了一种能有效降低复杂场景中虚碰撞干扰的生物启发式增强碰撞检测神经网络（ECDNN）。该网络包含突触前和突触后两部分神经子网络结构。其中，突触前子网络基于哺乳动物视网膜信息分层处理，逐级传递特性，从低阶视觉信息感知全局膨胀视焦点（FOE）来划分动态聚焦感受野，以此获取关键视觉信息。突触后子网络整合聚焦感受野内由迫近视觉刺激引发的膜电位兴奋响应，输出表征迫近碰撞危险的警报信号。实验表明，该模型不仅能有效过滤复杂场景中虚碰撞干扰，降低模型误检，还将碰撞检测准确率提升至96%以上，可为构建人工智能交互系统提供重要基础。

Abstract: Using traditional computer vision technology to resolve collision detection in complex scenes is a very difficult task, especially when faced with false collision interference, the model has a high false alarm rate and low accuracy. To address this problem, based on the hierarchical structure of the mammalian retina, this paper uses the danger perception characteristics of neurons in the Polysensory Zone (PZ) the precentral gyrus of the primate cerebral cortex to a specific visual area, and proposes a bio-inspired Enhanced Collision Detection Neural Network (ECDNN) that could effectively reduce false collision interference. This network consists of a presynaptic subnetwork and a postsynaptic subnetwork. Among them, the presynaptic subnetwork is based on the hierarchical processing and step-by-step transmission characteristics of mammalian retinal information, and divides the dynamic focus receptive field from the global Focus of Expansion (FOE) to obtain key visual information from low-order visual information perception. The postsynaptic subnetwork integrates the membrane potential excitation intensity caused by the approaching visual stimulus in the focus receptive field, and outputs an alarm signal representing the imminent collision danger. Experiments show that the model can not only effectively filter false collision interference in complex scenes and reduce model false detection, but also improve the accuracy of collision detection to over 96%, which can provide an important foundation for building future artificial intelligence interactive systems.

皮成东, 胡滨. 基于膨胀视焦点的生物启发碰撞检测神经网络[J]. 计算机工程, doi: 10.19678/j.issn.1000-3428.0252184.

PI Chengdong, HU Bin. Bio-Inspired Neural Network for Collision Detection Based on Focus of Expansion[J]. Computer Engineering, doi: 10.19678/j.issn.1000-3428.0252184.

参考文献

[1] JI X, ZHAO Y, LU S, et al. Robot path tracking and real-time obstacle avoidance under laser radar technology[C]//2023 IEEE 5th International Conference on Power, Intelligent Computing and Systems (ICPICS). Shenyang, China: IEEE, 2023: 512-518.
[2] LI H, KINSNER W, WANG Y, et al. Airborne radar based collision detection and avoidance system for unmanned aircraft systems in a varying environment[C]//2021 IEEE International Conference on Wireless for Space and Extreme Environments (WiSEE). Cleveland, USA: IEEE, 2021: 43-48.
[3] YASIN J N, MOHAMED S A S, HAGHBAYAN M H, et al. Low-cost ultrasonic based object detection and collision avoidance method for autonomous robots[J]. International Journal of Information Technology, 2021, 13: 97-107.
[4] YUE S, RIND F C. Collision detection in complex dynamic scenes using an LGMD-based visual neural network with feature enhancement[J]. IEEE Transactions on Neural Networks, 2006, 17(3): 705-716.
[5] LI T, ZHAO Z, ZHU M, et al. Cable vector collision detection algorithm for multi-robot collaborative towing system[J]. Journal of Shanghai Jiaotong University (Science), 2023: 1-11.
[6] WU J, LIAO Z, HAN Y, et al. An integral design for high performance sensor-less collision detection of serial robots[C]//2022 IEEE/ASME International Conference onAdvanced Intelligent Mechatronics (AIM). Sapporo, Japan: IEEE, 2022: 1714-1721.
[7] HU B, ZHANG Z. Bio-inspired visual neural network on spatio-temporal depth rotation perception[J]. Neural Computing and Applications, 2021, 33(16): 10351-10370.
[8] HU B, ZHANG Z. Bio-plausible visual neural network for spatio-temporally spiral motion Neurocomputing, 2018, 310: 96-114. perception[J].
[9] GRAZIANO M S A, COOKE D F. Parieto-frontal interactions, personal space, and defensive behavior[J]. Neuropsychologia, 2006, 44(6): 845-859.
[10] ZHAO J, XI S, LI Y, et al. A fly inspired solution to looming detection for collision avoidance[J]. IScience, 2023, 26(4): 106337.
[11] ZHANG Z, YUE S, ZHANG G. Fly visual system inspired artificial neural network for collision detection[J]. Neurocomputing, 2015, 153(8): 221-234.
[12] LI L, ZHANG Z, LU J. Artificial fly visual joint perception neural network inspired by multiple-regional collision detection[J]. Neural Networks, 2021, 135: 13-28.
[13] KATSAMPIRIS-SALGADO K, HANINGER K, GKRIZIS C, et al. Collision detection for collaborative assembly operations on high-payload robots[J]. Robotics and Computer-Integrated Manufacturing, 2024, 87: 102708.
[14] SHI C, DONG Z, PUNDLIK S, et al. A hardware-friendly optical flow-based time-to-collision algorithm[J]. Sensors, 2019, 19(4): 807. estimation
[15] XIN X, LIU K, YANG Z, et al. A probabilistic risk approach for the collision detection of multi-ships under spatiotemporal movement uncertainty[J]. Reliability Engineering & System Safety, 2021, 215: 107772.
[16] TRACY K, HOWELL T A, MANCHESTER Z. Differentiable collision detection for a set of convex primitives[C]//2023 IEEE International Conference on Robotics and Automation (ICRA). London, UK: IEEE, 2023: 3663-3670.
[17] MONTAUT L, LE LIDEC Q, BAMBADE A, et al. Differentiable collision detection: a randomized smoothing approach[C]//2023 IEEE International Conference on Robotics and Automation (ICRA). London, UK: IEEE, 2023: 3240-3246.
[18] HEIBERG A, LARSEN T N, MEYER E, et al. Risk-based implementation of COLREGs for autonomous surface vehicles using deep reinforcement learning[J]. Neural Networks, 2022, 152: 17-33.
[19] ALMUTAIRI M S, ALMUTAIRI K, CHIROMA H. Hybrid of deep recurrent network and long short term memory for rear-end collision detection in fog based internet of vehicles[J]. Expert Systems with Applications, 2023, 213: 119033.
[20] ZHOU B, LI Z, KIM S, et al. Shallow neural networks trained to detect collisions recover features of visual loom-selective neurons[J]. Elife, 2022, 11: e72067.
[21] RILL R A, FARAGÓ K B. Collision avoidance using deep learning-based monocular vision[J]. SN Computer Science, 2021, 2(5): 375.
[22] CHAN F H, CHEN Y T, XIANG Y, et al. Anticipating accidents in dashcam videos[C]//Computer Vision–ACCV 2016: 13th Asian Conference on Computer Vision. Cham, Switzerland: Springer, 2017: 136-153.
[23] BAO W, YU Q, KONG Y. Uncertainty-based traffic accident anticipation with spatio-temporal relational learning[C]//Proceedings of the 28th ACM International Conference on Multimedia. New York, USA: ACM, 2020: 2682-2690.
[24] FU Q, WANG H, HU C, et al. Towards computational models and applications of insect visual systems for motion perception: A review[J]. Artificial Life, 2019, 25(3): 263-311.
[25] YAKUBOWSKI J M, MCMILLAN G A, GRAY J R. Background visual motion affects responses of an insect motion-sensitive neuron to objects deviating from a collision course[J]. Physiological Reports, 2016, 4(10): e12801.
[26] QIN Z, FU Q, PENG J. A computationally efficient and robust looming perception model based on dynamic neural field[J]. Neural Networks, 2024, 179: 106502.
[27] ZHENG Y, WANG Y, WU G, et al. Enhancing LGMD-based model for collision prediction via binocular structure[J]. Frontiers in Neuroscience, 2023, 17: 1247227.
[28] CHANG Z, FU Q, CHEN H, et al. A look into feedback neural computation upon collision selectivity[J]. Neural Networks, 2023, 166: 22-37.
[29] CHANG Z, FU Q, HUA M, et al. Feedback neural computation in collision perception: Towards diverse selectivity[J]. Neurocomputing, 2024, 598: 128110.
[30] FU Q, HU C, PENG J, et al. Shaping the collision selectivity in a looming sensitive neuron model with parallel on and off pathways and spike frequency adaptation[J]. Neural Networks, 2018, 106: 127-143.
[31] LI J, SUN X, LI H, et al. On the ensemble of collision perception neuron models towards ultra-selectivity[C]//2023 International Joint Conference on Neural Networks (IJCNN). Gold Coast, Australia: IEEE, 2023: 1-8.
[32] FU Q, HU C, PENG J, et al. A robust collision perception visual neural network with specific selectivity to darker objects[J]. IEEE Transactions on Cybernetics, 2019, 50(12): 5074-5088.
[33] HONG J, FU Q, SUN X, et al. Boosting collision perception against noisy signals with a probabilistic neural network[C]//2023 International Joint Conference on Neural Networks (IJCNN). Gold Coast, Australia: IEEE, 2023: 1-8.
[34] CHANG Z, CHEN H, HUA M, et al. A bio-inspired visual collision detection network integrated with dynamic temporal variance feedback regulated by scalable functional countering jitter streaming[J]. Neural Networks, 2025, 182: 106882.
[35] 章盛, 郑胜男, 沈洁, 等. 果蝇嗅视神经通路研究综述 [J]. 电子与信息学报, 2024, 46(6): 2335-2351. ZHANG S, ZHENG S N, SHEN J, et al. Review on olfactory and visual neural pathways in drosophila[J]. Journal of Electronics & Information Technology, 2024, 46(6): 2335-2351.
[36] GRAZIANO M S A, HU X T, GROSS C G. Visuospatial properties of ventral premotor cortex[J]. Journal of Neurophysiology, 1997, 77(5): 2268-2292.
[37] WANG X Y, GONG X M, SUN Q, et al. Attractive effects of previous form information on heading estimation from optic flow occur at perceptual stage[J]. Journal of Vision, 2022, 22(12): 18.
[38] STRONG S L, SILSON E H, GOUWS A D, et al. Differential processing of the direction and focus of expansion of optic flow stimuli in areas MST and V3A of the human visual cortex[J]. Journal of Neurophysiology, 2017, 117(6): 2209-2217.
[39] HUANG R, ERICSON S. An efficient way to estimate the focus of expansion[C]//2018 IEEE 3rd International Conference on Image, Vision and Computing (ICIVC). Chongqing, China: IEEE, 2018: 691-695.
[40] HAHN J, MONAVARFESHANI A, QIAO M, et al. Evolution of neuronal cell classes and types in the vertebrate retina[J]. Nature, 2023, 624(7991): 415-424.
[41] SHEKHAR K, LAPAN S W, WHITNEY I E, et al. Comprehensive classification of retinal bipolar neurons by single-cell transcriptomics[J]. 1308-1323. Cell, 2016, 166(5):
[42] 赵轻轻, 胡滨. 不变性全局稀疏轮廓点表征的运动行人检测神经网络[J]. 计算机应用, 2025, 45(4): 1271-1284. ZHAO Q Q, HU B. Moving pedestrian detection neural network with invariant global sparse contour point representation[J]. Journal of Computer Applications, 2025, 45(4): 1271-1284.
[43] VLASITS A L, EULER T, FRANKE K. Function first: classifying cell types and circuits of the retina[J]. Current Opinion in Neurobiology, 2019, 56: 8-15.
[44] WEST E R, CEPKO C L. Development and diversification of bipolar interneurons in the mammalian retina[J]. Developmental Biology, 2022, 481: 30-42.
[45] 于世海, 胡滨. 生物启发的运动行人视觉不变性响应神经网络[J/OL]. 计算机科学 : 1-22[2025-03-01]. http://kns.cnki.net/kcms/detail/50.1075.TP.20240913.0915. 006.html. YU S H, HU B. Bio-inspired neural network with visual invariant response to moving pedestrian[J/OL]. Computer Science: 1-22[2025-03-01]. http://kns.cnki.net/kcms/detail/50.1075.TP.20240913.0915 .006.html.
[46] CALONDER M, LEPETIT V, STRECHA C, et al. Brief: Binary robust independent elementary features[C]//Computer Vision–ECCV 2010: 11th European Conference on Computer Vision. Berlin, Germany: Springer, 2010: 778-792.
[47] SHAH A P, LAMARE J B, NGUYEN-ANH T, et al. CADP: A novel dataset for CCTV traffic camera based accident analysis[C]//2018 15th IEEE International Conference on Advanced Video and Signal Based Surveillance (AVSS). Auckland, New Zealand: IEEE, 2018: 1-9.
[48] MOURA D C, ZHU S, ZVITIA O. Nexar dashcam collision prediction dataset and challenge[J]. arXiv preprint arXiv:2503.03848, 2025.
[49] GEIGER A, LENZ P, URTASUN R. Are we ready for autonomous driving? the kitti vision benchmark suite[C]//2012 IEEE Conference on Computer Vision and Pattern Recognition. Providence, USA: IEEE, 2012: 3354-3361.
[50] TEMIZER I, DONOVAN J C, BAIER H, et al. A visual pathway for looming-evoked escape in larval zebrafish[J]. Current Biology, 2015, 25(14): 1823-1834.
[51] WANG Y, GONG Y, HUANG S, et al. Memristor-based biomimetic compound eye for real-time collision detection[J]. Nature Communications, 2021, 12(1): 5979.

选择文件类型/文献管理软件名称

选择包含的内容