Sophisticated deep learning with on-chip optical diffractive tensor processing

Ever-growing deep-learning technologies are making revolutionary changes for modern life.However,conventional computing architectures are designed to process sequential and digital programs but are burdened with performing massive parallel and adaptive deep-learning applications.Photonic integrated circuits provide an efficient approach to mitigate bandwidth limitations and the power-wall brought on by its electronic counterparts,showing great potential in ultrafast and energy-free high-performance computation.Here,we propose an optical computing architecture enabled by on-chip diffraction to implement convolutional acceleration,termed“optical convolution unit”(OCU).We demonstrate that any real-valued convolution kernels can be exploited by the OCU with a prominent computational throughput boosting via the concept of structral reparameterization.With the OCU as the fundamental unit,we build an optical convolutional neural network(oCNN)to implement two popular deep learning tasks:classification and regression.For classification,Fashion Modified National Institute of Standards and Technology(Fashion-MNIST)and Canadian Institute for Advanced Research(CIFAR-4)data sets are tested with accuracies of 91.63%and 86.25%,respectively.For regression,we build an optical denoising convolutional neural network to handle Gaussian noise in gray-scale images with noise levelσ=10,15,and 20,resulting in clean images with an average peak signal-to-noise ratio(PSNR)of 31.70,29.39,and 27.72 dB,respectively.The proposed OCU presents remarkable performance of low energy consumption and high information density due to its fully passive nature and compact footprint,providing a parallel while lightweight solution for future compute-in-memory architecture to handle high dimensional tensors in deep learning.

Integrated diffractive optical neural network with space-time interleaving

Integrated diffractive optical neural networks(DONNs)have significant potential for complex machine learning tasks with high speed and ultralow energy consumption.However,the on-chip implementation of a high-performance optical neural network is limited by input dimensions.In contrast to existing photonic neural networks,a space-time interleaving technology based on arrayed waveguides is designed to realize an on-chip DONN with high-speed,high-dimensional,and all-optical input signal modulation.To demonstrate the performance of the on-chip DONN with high-speed space-time interleaving modulation,an on-chip DONN with a designed footprint of 0.0945 mm~2is proposed to resolve the vowel recognition task,reaching a computation speed of about 1.4×10^(13)operations per second and yielding an accuracy of 98.3%in numerical calculation.In addition,the function of the specially designed arrayed waveguides for realizing parallel signal inputs using space-time conversion has been verified experimentally.This method can realize the on-chip DONN with higher input dimension and lower energy consumption.

Robust Identification of Road Surface Condition Based on Ego‑Vehicle Trajectory Reckoning

The type of road surface condition(RSC)will directly affect the driving performance of vehicles.Monitoring the type of RSC is essential for both transportation agencies and individual drivers.However,most existing methods are solely based on a dynamics-based method or an image-based method,which is susceptible to road excitation limitations and interference from the external environment.Therefore,this paper proposes a decision-level fusion identification framework of the RSC based on ego-vehicle trajectory reckoning to accurately obtain the type of RSC that the front wheels of the vehicle will expe-rience.First,a road feature extraction model based on multi-task learning is conducted,which can simultaneously segment the drivable area and road cast shadow.Second,the optimized candidate regions of interest are classified with confidence levels by ShuffleNet.Considering environmental interference,candidate regions of interest regarded as virtual sensors are fused by improved Dempster-Shafer evidence theory to obtain the fusion results.Finally,the ego-vehicle trajectory reckoning module based on the kinematic bicycle model is added to the proposed fusion method to extract the RSC experienced by the front wheels.The performance of the entire framework is verified on a specific dataset with shadow and split curve roads.The results reveal that the proposed method can identify the RSC with accurate predictions in real time.

Mass estimation method for intelligent vehicles based on fusion of machine learning and vehicle dynamic model

Vehicle mass is an important parameter for motion control of intelligent vehicles,but is hard to directly measure using normal sensors.Therefore,accurate estimation of vehicle mass becomes crucial.In this paper,a vehicle mass estimation method based on fusion of machine learning and vehicle dynamic model is introduced.In machine learning method,a feedforward neural network(FFNN)is used to learn the relationship between vehicle mass and other state parameters,namely longitudinal speed and acceleration,driving or braking torque,and wheel angular speed.In dynamics-based method,recursive least square(RLS)with forgetting factor based on vehicle dynamic model is used to estimate the vehicle mass.According to the reliability of each method under different conditions,these two methods are fused using fuzzy logic.Simulation tests under New European Driving Cycle(NEDC)condition are carried out.The simulation results show that the estimation accuracy of the fusion method is around 97%,and that the fusion method performs better stability and robustness compared with each single method.