Koninklijke Philips Corporation N.v.

Michigan State University

Philips Electronics N.a. Corporation

Ford Global Technolgoies, LLC

U.S. Philips Corporation

Electronics and Telecommunications Research Institute

Philips Electronics North America Corporation

Koninklijke Phillips Electronics N.v.

Hayder Radha

Mihaela Van Der Schaar

Yingwei Chen

Robert A Cohen

Kavitha Parthasarathy

Su Pang

Mahesh Balakrishnan

Daniel Morris

Dmitri Loguinov

Jin-Woo Hong

Mark Gehrke

Jeong-Ju Yoo

Mihaela Van Der Schaar-Mitrea

Yong-Ju Cho

Hothaifa Al-Qassab

Shirish Krande

Kiran Misra

Abo Talib Mahfoodh

Mazin Hnewa

David Conger

Jon Diedrich

A method and system of operating a vehicle includes a first sensor generating first sensor data for an object comprising a first bounding box from a first sensor. The first sensor data comprising a first confidence score. A second sensor generates second sensor data for the object comprising a second bounding box from a second sensor different than the second sensor. The second sensor data comprises a second confidence score. A bounding box circuit is programmed to generate a third confidence score for the object based on the first sensor data and the second sensor data and utilize the first sensor data, the second sensor data and the third confidence score to control operation of a vehicle system.

Lidar camera fusion for autonomous vehicles

A method and system for multi-object tracking is set forth. Object data for boundaries of a plurality of objects are received. Poisson multi-Bernoulli mixture filtering is performed on the object data to form a filtered set of object data. Ultimately, the filtered set of object data is used to control the operation of the vehicle. Identifiers and probabilities are associated with the objects to reduce the set of object data.

Multi-object tracking for autonomous vehicles

Accurate vehicle localization is arguably the most critical and fundamental task for autonomous vehicle navigation. While dense 3D point-cloud-based maps enable precise localization, they impose significant storage and transmission burdens when used in city-scale environments. A highly compressed representation for LiDAR maps, along with an efficient and robust real-time alignment algorithm for on-vehicle LiDAR scans, is proposed here. The proposed mapping framework, requires less than 0.1% of the storage space of the original 3D point cloud map. In essence, mapping framework emulates an original map through feature likelihood functions. In particular, the mapping framework models planar, pole and curb features. These three feature classes are long-term stable, distinct and common among vehicular roadways. Multiclass feature points are extracted from LiDAR scans through feature detection. A new multiclass-based point-to-distribution alignment method is also proposed to find the association and alignment between the multiclass feature points and the map.

Mapping and localization system for autonomous vehicles

Advanced automotive active-safety systems, in general, and autonomous vehicles, in particular, rely heavily on visual data to classify and localize objects, most notably pedestrians and other nearby cars, to assist the corresponding vehicles maneuver safely in their environment. However, the performance of object detection methods is anticipated to degrade under challenging rainy conditions. Nevertheless, and despite major advancements in the development of deraining approaches, the impact of rain on object detection has largely been understudied, especially in the context of autonomous systems. This disclosure analyzes this problem space and presents an improved system for detecting objects under rainy conditions.

Object detection under rainy conditions for autonomous systems

By exchanging basic traffic messages among vehicles for safety applications, a significantly higher level of safety can be achieved when vehicles and designated infrastructure-locations share their sensor data. While cameras installed in one vehicle can provide visual information for mitigating many avoidable accidents, a new safety paradigm is envisioned where visual data captured by multiple vehicles are shared and fused for significantly more optimized active safety and driver assistance systems. The sharing of visual data is motivated by the fact that some critical visual views captured by one vehicle or by an infrastructure-location are not visible or captured by other vehicles in the same environment. Sharing such data in real-time provides an invaluable new level of awareness that can significantly enhance a driver-assistance, connected vehicle, and/or autonomous vehicle's safety-system.

Visual sensor fusion and data sharing across connected vehicles for active safety

The streaming media encoding and routing system employs an encoder circuit that constructs a streaming media sequence as a plurality of sequential frames, each frame comprising a plurality of segments. The encoder circuit has a processor that is programmed to place media information into the plurality of segments of each frame according to a predefined priority and further programmed to order the sequence of said segments within each frame such that the segments are progressively priority-ranked from high priority to low priority to define an EPIQ-encoded packet. A processor tests an incoming packet to determine whether it is EPIQ-encoded. The processor reduces network congestion when the received quality of service data indicates presence of network congestion by: (a) selectively and progressively erasing segments within an incoming packet beginning with a lowest-ranked priority segment if the incoming packet is determined to be EPIQ-encoded; and (b) dropping the entire incoming packet if the incoming packet is determined not be EPIQ-encoded.

Streaming media using erasable packets within internet queues

A computer-implemented method is provided for compressing a collection of images. The method includes: receiving image data for an image ensemble composed of a plurality of images; dividing the image data into a plurality of data blocks; decomposing each data block from the plurality of data blocks into a set of rank-one tensors by estimating each rank-one tensor individually, where each rank-one tensor is represented by three one-dimensional vectors; arranging the rank-one tensors for each of the plurality of data blocks into a two-dimensional matrix, where the decomposer is implemented by computer processor; compressing the two-dimensional matrix using a compression operation; and storing the compressed two-dimensional matrix in a non-transitory data store.

Compression of image ensembles using tensor decomposition

A computer-implemented method for resizing an image using a seam carving algorithm. The method may include measuring energy levels of pixels in an original image to derive an original energy map; applying a filter to an original energy map to derive a first energy map having a scale less than the original energy map; iteratively applying the filter N times, starting with the first energy map, to an energy map from an immediately preceding iteration; upsampling each of the energy maps to a resolution that matches the original energy map; combining the upsampled energy maps with the original energy map to form a composite image; identifying a seam by finding a path in the composite image having lowest energy quantities from one end of the composite image to an opposing end of the composite image; and selectively deleting the identified seam from the original image, thereby yielding a resized image.

Generalized multiscale seam carving framework

There is provided a channel capacity estimation method for adaptive video transmission, the method including: estimating a Bit Error Rate (BER) for a received video packet; and estimating channel capacity by using the estimated BER.

Channel capacity estimation and prediction method and apparatus for rate adaptive wireless video

Improved efficiency in transmission of FGS encoded video data is achieved by reducing the transmission of overhead information items necessary to decode the received video data. After establishing an initial or first set of criteria for selectively enhancing areas within an image, each subsequent transmission frame includes an indicator that informs a receiving system to decode the current frame using the previously established criteria. The initial criteria include shift factor values specifying selectively enhanced areas of an image; position, displacement vector, size and enhancement factor for at least one area of interest with in an image; and known fixed values contained at both the transmitting and receiving systems.

Method and apparatus for improved efficiency in transmission of fine granular scalable selective enhanced images

A system and method are disclosed for streaming scalable video data over a variable-bandwidth network such as a packet-based one. In other words, the number of bits (for FGS) or sub-layers (for discrete multi-layer scalability) is chosen to be transmitted for each enhancement layer frame. Further, these bits or sub-layers are also partitioned over multiple packets if being sent over a packet-based network. Further, a windowing function is used to smooth the variation in the number of bits or sub-layers transmitted for each frame in situations where the network bandwidth is varying.

Method and apparatus for streaming scalable video

A video coding technique having motion compensation for bi-directional predicted frames (B-frames) and predicted frames (P-frames). The video coding technique involves a single-loop prediction-based base layer which includes base layer B- and/or P-frames that are generated from 'extended' or 'enhanced' base layer reference frames during base layer coding.

Single-loop motion-compensation fine granular scalability

A video coding technique having motion compensation within a fine granular scalable coded enhancement layer. In one embodiment, the video coding technique involves a two-loop prediction-based enhancement layer including non-motion-predicted enhancement layer I- and P-frames and motion-predicted enhancement layer B-frames. The motion-predicted enhancement layer B-frames are computed using: 1) motion-prediction from two temporally adjacent differential I- and P- or P- and P-frame residuals, and 2) the differential B-frame residuals obtained by subtracting the decoded base layer B-frame residuals from the original base layer B-frame residuals. In a second embodiment, the enhancement layer further includes motion-predicted enhancement layer P-frames. The motion-predicted enhancement layer P-frames are computed using: 1) motion-prediction from a temporally adjacent differential I- or P-frame residual, and 2) the differential P-frame residual obtained by subtracting the decoded base layer P-frame residual from the original base layer P-frame residual.

Double-loop motion-compensation fine granular scalability

Disclosed is a system and method for estimating retransmission timeout (RTO) in a real-time streaming applications over the Internet between a server and a client. Accordingly, the present invention employs retransmission timeout (RTO) in NACK-based applications to support multiple retransmission attempts per lost packet, wherein the RTO is estimated by an actual around-trip delay (RTT) and a smooth inter-packet delay variance.

Method for efficient retransmission timeout estimation in NACK-based protocols

The present invention is directed to a method for encoding video data in an embedded fashion in order to achieve fine granular scalable video. The method includes the video data being transformed into a plurality of DCT coefficients. Further, the DCT coefficients are arranged into sub-groups and the DCT coefficients are scanned according to the sub-groups. The DCT coefficients being scanned by the sub-groups enables a higher level of scalability to be achieved.

Growing community of inventors

Mahwah, NJ, United States of America

Hayder Radha