Abstract: Disclosed is a method and device for processing data, and the method includes generating a target augmentation task sequence by processing the target data with a trained first model that performs inference on the target data to generate the target data augmentation task sequence, generate augmented target data by performing data augmentation on the target data according to the target augmentation task sequence, and obtaining a prediction result corresponding to the target data by inputting the augmented target data to a trained second model and performing a corresponding processing on the augmented target data by the trained second model.

Abstract: The disclosure discloses a method for detecting an intrusion in parallel based on an unbalanced data Deep Belief Network, which reads an unbalanced data set DS; under-samples the unbalanced data set using the improved NCR algorithm to reduce the ratio of the majority type samples and make the data distribution of the data set balanced; the improved differential evolution algorithm is used on the distributed memory computing platform Spark to optimize the parameters of the deep belief network model to obtain the optimal model parameters; extract the feature of data of the data set, and then classify the intrusion detection by the weighted nuclear extreme learning machine, and finally train multiple weighted nuclear extreme learning machines of different structures in parallel by multithreading as the base classifier, and establish a multi-classifier intrusion detection model based on adaptive weighted voting for detecting the intrusion in parallel.

Abstract: System and method for training a machine learning model are disclosed. In one embodiment, for each of the driving scenarios, responsive to sensor data from one or more sensors of a vehicle and the driving scenario, driving statistics and environment data of the vehicle are collected while the vehicle is driven by a human driver in accordance with the driving scenario. Upon completion of the driving scenario, the driver is requested to select a label for the completed driving scenario and the selected label is stored responsive to the driver selection. Features are extracted from the driving statistics and the environment data based on predetermined criteria. The extracted features include some of the driving statistics and some of the environment data collected at the different points in time during the driving scenario.

Abstract: In an embodiment, a learning-based dynamic modeling method is provided for use with an autonomous driving vehicle. A control module in the ADV can generate current states of the ADV and control commands for a first driving cycle, and send the current states and control commands to a dynamic model implemented using a trained neural network model. Based on the current states and the control commands, the dynamic model generates expected future states for a second driving cycle, during which the control module generates actual future states. The ADV compares the expected future states and the actual future states to generate a comparison result, for use in evaluating one or more of a decision module, a planning module and a control module in the ADV.

Abstract: An autonomous driving system includes a number of sensors and a number of autonomous driving modules. The autonomous driving system further includes a global store to store data generated and used by processing modules such as sensors and/or autonomous driving modules. The autonomous driving system further includes a task scheduler coupled to the sensors, the autonomous driving modules, and the global store. In response to output data generated by any one or more of processing modules, the task scheduler stores the output data in the global store. In response to a request from any of the processing modules for processing data, the task scheduler provides input data stored in the global store to the processing module. The task scheduler is executed in a single thread that is responsible for managing data stored in the global store and dispatching tasks to be performed by the processing modules.

Abstract: An event queue is maintained to store IO events generated from a number of sensors and timer events generated for a number of autonomous driving modules. For each of the events pending in the event queue, in response to determining that the event is an IO event, the data associated with the IO event is stored in a data structure associated with the sensor in a global store. In response to determining that the event is a timer event, a worker thread associated with the timer event is launched. The worker thread executes one of the autonomous driving modules triggered or initiated the timer event. Input data is retrieved from the global store and provided to the worker thread to allow the worker thread to process the input data.

Abstract: In an embodiment, a learning-based dynamic modeling method is provided for use with an autonomous driving vehicle. A control module in the ADV can generate current states of the ADV and control commands for a first driving cycle, and send the current states and control commands to a dynamic model implemented using a trained neural network model. Based on the current states and the control commands, the dynamic model generates expected future states for a second driving cycle, during which the control module generates actual future states. The ADV compares the expected future states and the actual future states to generate a comparison result, for use in evaluating one or more of a decision module, a planning module and a control module in the ADV.

Abstract: A global store is maintained to store a number of data structures. Each data structure includes a number of entries and each entry stores data of at least one event in a chronological order. Each data structure is associated with at least one sensor or an autonomous driving module of an autonomous driving vehicle. When a first event associated with a first autonomous driving module is received, where the first event includes a first topic ID, the first topic ID is hashed to identify a first data structure corresponding to the first event. A pointer pointing to a head of the first data structure is passed to the first autonomous driving module to allow the first autonomous driving module to process data associated with the first event.

Abstract: System and method for training a machine learning model are disclosed. In one embodiment, for each of the driving scenarios, responsive to sensor data from one or more sensors of a vehicle and the driving scenario, driving statistics and environment data of the vehicle are collected while the vehicle is driven by a human driver in accordance with the driving scenario. Upon completion of the driving scenario, the driver is requested to select a label for the completed driving scenario and the selected label is stored responsive to the driver selection. Features are extracted from the driving statistics and the environment data based on predetermined criteria. The extracted features include some of the driving statistics and some of the environment data collected at the different points in time during the driving scenario.

Abstract: An autonomous driving system includes a number of sensors and a number of autonomous driving modules. The autonomous driving system further includes a global store to store data generated and used by processing modules such as sensors and/or autonomous driving modules. The autonomous driving system further includes a task scheduler coupled to the sensors, the autonomous driving modules, and the global store. In response to output data generated by any one or more of processing modules, the task scheduler stores the output data in the global store. In response to a request from any of the processing modules for processing data, the task scheduler provides input data stored in the global store to the processing module. The task scheduler is executed in a single thread that is responsible for managing data stored in the global store and dispatching tasks to be performed by the processing modules.

Abstract: A global store is maintained to store a number of data structures. Each data structure includes a number of entries and each entry stores data of one of the events in a chronological order. Each data structure is associated with one of the sensors or the autonomous driving modules of an autonomous driving vehicle. When a first event associated with a first autonomous driving module is received, where the first event includes a first topic ID, the first topic ID is hashed to identify a first data structure corresponding to the first event. A pointer pointing to a head of the first data structure is passed to the first autonomous driving module to allow the first autonomous driving module to process data associated with the first event.

Abstract: An event queue is maintained to store IO events generated from a number of sensors and timer events generated for a number of autonomous driving modules. For each of the events pending in the event queue, in response to determining that the event is an IO event, the data associated with the IO event is stored in a data structure associated with the sensor in a global store. In response to determining that the event is a timer event, a worker thread associated with the timer event is launched. The worker thread executes one of the autonomous driving modules triggered or initiated the timer event. Input data is retrieved from the global store and provided to the worker thread to allow the worker thread to process the input data.

Abstract: Described is a system (and method) for generating a driving scenario for an autonomous driving simulator. The system may use a camera mounted to a vehicle as a cost effective approach to obtain real-life driving scenario data. The system may then analyze the two-dimensional image data to create a three-dimensional driving simulation. The analysis may include detecting objects (e.g. vehicles, pedestrians, etc.) within the two-dimensional image data and determining movements of the object based on a position, trajectory, and velocity of the object. The determined information of the object may then be projected onto a map that may be used for generating the three-dimensional driving simulation. The use of cost-effective cameras provides the ability to obtain vast amounts of driving image data that may be used to provide an extensive coverage of the potential types of driving scenarios an autonomous vehicle may encounter.