Abstract: The invention relates to an automated driving control system, including: a data exchange unit configured to obtain video data streams and distribute the video data streams to at least one computing unit; the at least one computing unit configured to compute perception result data based on the video data streams; a sensor fusion unit configured to fuse the perception result data and sensor data, to obtain fusion result data; and a planning control unit configured to generate a driving control instruction based on the fusion result data, where the planning control unit or the sensor fusion unit is configured to provide a bypass of the data exchange unit. When anomalies occur in any of the units, the automated driving control system may provide a corresponding bypass to keep an automated driving function going.

Abstract: Provided are methods for a track refinement network. In examples, center boxes are obtained from a record of driving data, wherein a center box is a center of a sequence of boxes along a track, and the track is associated with a tracked object detected within the sequence of boxes, each respective box comprising a center, a size, and an orientation. Track windows are generated around respective center boxes, wherein a track window corresponds to a respective center box along the track. Track windows are cropped and normalized with respect to center boxes to enable single refinement model for multiple object classes. Point cloud features and trajectory features are extracted from the cropped and normalized track windows. The point cloud features and trajectory features are input into a track refinement network, wherein the track refinement network uses features from the entire track to output a refined center, a refined size, and a refined orientation of each respective center box.

Abstract: Provided are methods for track segment cleaning of tracked objects using neural networks, which can include detecting a first track segment and a second track segment. The method includes applying a machine learning model trained to determine if the first track segment and second track segment capture real objects and if the first track segment and the second track segment are representative of an identical object exterior to a vehicle. The method further includes combining the first track segment and the second track segment to form a single track segment having a single trajectory in response to the first track segment and the second track segment being determined to be representative of the identical object. Systems and computer program products are also provided.

Abstract: Provided is an electrohydrostatic actuation system including an emergency shut-off circuit to be actuated stably with a simple configuration. The electrohydrostatic actuation system includes: a hydraulic cylinder (24) including a piston (25) to which a valve element is connected, a first chamber (24A), and a second chamber (24B); a hydraulic pump (21) configured to supply hydraulic fluid to the first chamber (24A) or the second chamber (24B); a servo motor (M) configured to drive the hydraulic pump (21); a shuttle valve (11) configured to establish communication to a downstream side under a state in which a hydraulic pressure generated by the hydraulic pump (21) is maintained; a solenoid valve (12) configured to receive the hydraulic pressure via the shuttle valve (11) as a pilot pressure; and a logic valve (13) including a first port configured to receive the pilot pressure from the solenoid valve (12), and a second port to be communicated to the first chamber (24A) of the hydraulic cylinder (24).

Abstract: Provided are methods for track segment cleaning of tracked objects using neural networks, which can include detecting a first track segment and a second track segment. The method includes applying a machine learning model trained to determine if the first track segment and second track segment capture real objects and if the first track segment and the second track segment are representative of an identical object exterior to a vehicle. The method further includes combining the first track segment and the second track segment to form a single track segment having a single trajectory in response to the first track segment and the second track segment being determined to be representative of the identical object. Systems and computer program products are also provided.

Abstract: Among other things, techniques are described for spatially and temporally consistent ground modelling with information fusion. A vehicle location and orientation is obtained and an instantaneous ground height estimation is obtained for anchor cells of a grid map of the vehicle based on the obtained vehicle location and orientation. A pseudo ground height associated with non-anchor cells of the grid map is computed. A Bayesian filtering framework is used to generate a final ground height estimate for the anchor cells and the non-anchor cells based on the instantaneous ground height estimation for the anchor cells, the pseudo ground height associated with non-anchor cells, an estimated ground height from a previous timestamp, or any combinations thereof. The vehicle operates according to the final ground height estimate.

Abstract: Systems and methods are described for monitoring detection of objects in a sensor. The system can generate an occupancy array based on scene data corresponding to a scene of a vehicle. The array can include points that represent a location within the environment of the vehicle and an occupancy value that indicates whether an object is detected at that location. The system can modify an occupancy value of at least one point in the array and identify at least one static object miss based on a group of occupancy values of a group of points in the array.

Abstract: This disclosure describes systems and methods for training a neural network with a training data set including data items labeled at different granularities. During training, each item within the training data set can be fed through the neural network. For items with labels of a higher granularity, weights of the network can be adjusted based on a comparison between the output of the network and the label of the item. For items with labels of a lower granularity, an output of the network can be fed through a conversion function that convers the output from the higher granularity to the lower granularity. The weights of the network can then be adjusted based on a comparison between the converted output and the label of the item.

Abstract: Provided are methods for scalable and realistic camera blockage dataset generation, which can include generating synthetic images depicting a blockage on or near an imaging sensor. The synthetic images may be created by combining one or more chroma key-extracted partial blockage image with one or more background images, the combination of which can provide a scalable blockage dataset. Metadata for each synthetic image can be generated along with the synthetic image, by annotating the portion of the synthetic image represented by the chroma key-extracted partial blockage image as constituting blockage. The synthetic images can be used to increase the accuracy of machine learning models trained to identify blockage by increasing the volume of data available for such training.

Abstract: Provided are methods for learning to identify safety-critical scenarios for autonomous vehicles. First state information representing a first state of a driving scenario is received. The information includes a state of a vehicle and a state of an agent in the vehicle's environment. The first state information is processed with a neural network to determine at least one action to be performed by the agent, including a perception degradation action causing misperception of the agent by a perception system of the vehicle. Second state information representing a second state of the driving scenario is received after performance of the at least one action. A reward for the action is determined. First and second distances between the vehicle and the agent are determined and compared to determine the reward for the at least one action. At least one weight of the neural network is adjusted based on the reward.

Abstract: Provided is an electrohydrostatic actuation system including an emergency shut-off circuit to be actuated stably with a simple configuration. The electrohydrostatic actuation system includes: a hydraulic cylinder (24) including a piston (25) to which a valve element is connected, a first chamber (24A), and a second chamber (24B); a hydraulic pump (21) configured to supply hydraulic fluid to the first chamber (24A) or the second chamber (24B); a servo motor (M) configured to drive the hydraulic pump (21); a shuttle valve (11) configured to establish communication to a downstream side under a state in which a hydraulic pressure generated by the hydraulic pump (21) is maintained; a solenoid valve (12) configured to receive the hydraulic pressure via the shuttle valve (11) as a pilot pressure; and a logic valve (13) including a first port configured to receive the pilot pressure from the solenoid valve (12), and a second port to be communicated to the first chamber (24A) of the hydraulic cylinder (24).

Abstract: A method for the synthesis of diazabicyclo[6.2.0]decane compounds is provided. The synthesis proceeds by stereoselective synthesis of a chiral lactone followed by azetidine formation via a series of chemoselective reactions. Bicyclization results with the formation of diazobicyclo[6.2.0]decane related compounds.

Abstract: This disclosure describes the use of optical sensors to detect and characterize the state of traffic lights to assist with the navigation of autonomous vehicles. In particular, a specific optical configuration is shown that includes both a fixed-exposure sensor and an auto-exposure sensor. Imagery from the two sensor types can be combined to more accurately characterize the state of traffic signals at any particular intersection. Systems and methods for analyzing only select regions of the imagery captured by the traffic light detection system are also described.

Abstract: The disclosure relates to a p-type skutterudite material and a method of making the same, comprising providing a p-type skutterudite material having a general formula: IyFe4-xMxSb12/z(J) wherein I represents one or more filling atoms in a skutterudite phase, the total filling amount y satisfying 0.01?y?1; M represents one or more dopant atoms with the doping amount x satisfying 0?x?4; J represents one or more second phases with the molar ratio z satisfying 0?z?0.5; wherein second phase precipitates are dispersed throughout the skutterudite phase.

Abstract: A thermoelectric material includes a multiple transition metal-doped type I clathrate crystal structure having the formula A8TMy11TMy22 . . . TMynnMzX46-y1-y2- . . . -yn-z. In the formula, A is selected from the group consisting of barium, strontium, and europium; X is selected from the group consisting of silicon, germanium, and tin; M is selected from the group consisting of aluminum, gallium, and indium; TM1, TM2, and TMn are independently selected from the group consisting of 3d, 4d, and 5d transition metals; and y1, y2, yn and Z are actual compositions of TM1, TM2, TMn, and M, respectively. The actual compositions are based upon nominal compositions derived from the following equation: z=8·qA?|?q1|y1?|?q2|y2? . . . ?|?qn|yn, wherein qA is a charge state of A, and wherein ?q1, ?q2, ?qn are, respectively, the nominal charge state of the first, second, and n-th TM.

Abstract: A thermoelectric material includes a multiple transition metal-doped type I clathrate crystal structure having the formula A8TMy11TMy22 . . . TMynnMzX46-y1-y2- . . . -yn-z. In the formula, A is selected from the group consisting of barium, strontium, and europium; X is selected from the group consisting of silicon, germanium, and tin; M is selected from the group consisting of aluminum, gallium, and indium; TM1, TM2, and TMn are independently selected from the group consisting of 3d, 4d, and 5d transition metals; and y1, y2, yn and Z are actual compositions of TM1, TM2, TMn, and M, respectively. The actual compositions are based upon nominal compositions derived from the following equation: z=8·qA?|?q1|y1?|?q2|y2? . . . ?|?qn|yn, wherein qA is a charge state of A, and wherein ?q1, ?q2, ?qn are, respectively, the nominal charge state of the first, second, and n-th TM.

Abstract: The present invention provides system and method of clustering data from a data matrix. The method includes generating at least one initial cluster from the data matrix to form a submatrix and adding or removing a row or a column to reduce the average residue of the submatrix. The system includes means for generating at least one initial cluster from the data matrix to form a submatrix and means for adding or removing a row or a column to reduce the average residue of the submatrix.

Abstract: Periodic patterns in time series data can be hierarchical in nature, where a higher level pattern may comprise repetitions of lower level patterns. In the presence of noises, these repetitions of lower level patterns may not be perfect. A novel model, namely a meta-pattern, is provided in accordance with the present invention to capture these higher level patterns. The meta-pattern can not only provide a more compact representation of patterns but also capture the regularities of pattern evolutions, which may not be expressed by previous models due to the presence of noise. A method is provided to mine meta-patterns in an iterative manner by discovering meta-patterns and their supporting subsequences in the form of lists of segments of contiguous repetitions of a meta-pattern. The number of pattern repetitions in each said segment is at least a predefined threshold min_rep and the distance between any two adjacent segments is at most a predefined threshold max_dis.

Abstract: A system and method are provided for discovering significant patterns from a list of records in a dataset. Each record includes a set of items, and each significant pattern includes a subset of items such that a significance of the pattern exceeds a significance level. A significance is computed for each item in the list of records to determine significant items. The records are randomly sampled to select a sample portion of the records. Ambiguous patterns are identified against the sample portion of the records and verified against the entire list of records in the dataset.