A deep neural network(s) (DNN) may be used to detect objects from sensor data of a three dimensional (3D) environment. For example, a multi-view perception DNN may include multiple constituent DNNs or stages chained together that sequentially process different views of the 3D environment. An example DNN may include a first stage that performs class segmentation in a first view (e.g., perspective view) and a second stage that performs class segmentation and/or regresses instance geometry in a second view (e.g., top-down). The DNN outputs may be processed to generate 2D and/or 3D bounding boxes and class labels for detected objects in the 3D environment. As such, the techniques described herein may be used to detect and classify animate objects and/or parts of an environment, and these detections and classifications may be provided to an autonomous vehicle drive stack to enable safe planning and control of the autonomous vehicle.

Driver assistance systems for detecting a structural barrier extending along a road. The driver assistance system may be mountable in a host vehicle. The camera may capture multiple image frames in the forward field of view of the camera. A processor may process motion of images of the barrier in the image frames. The camera may be a single camera. The motion of the images may be responsive to forward motion of the host vehicle and/or the motion of the images may be responsive to lateral motion of the host vehicle.

A vehicular vision system includes a camera disposed at an in-cabin side of a windshield of a vehicle and viewing forward of the vehicle. The control, responsive at least in part to image processing by an image processor of multiple frames of captured image data, detects an object present exterior of the equipped vehicle that is moving relative to the equipped vehicle. The control receives vehicle motion data indicative of motion of the vehicle when the vehicle is moving. The control, responsive at least in part to the received vehicle motion data, and via image processing of multiple frames of captured image data, determines motion of the detected object relative to the moving vehicle by (i) determining corresponding object points in at least two frames of captured image data and (ii) estimating object motion trajectory of the detected object based at least in part on the determined corresponding object points.

A method for controlling a movable object includes obtaining current location information of an obstacle while the movable object tracks a target, determining whether the obstacle is located in a reactive region relative to the movable object based on the current location information of the obstacle, and, in response to determining that the obstacle is not located in the reactive region, selecting an optimized set of candidate movement characteristics having an optimized route optimization score from multiple sets of candidate movement characteristics, and adjusting one or more movement characteristics of the movable object based on the optimized set of candidate movement characteristics such that a distance between the movable object and the obstacle is maintained at or beyond a predefined distance. Each set of candidate movement characteristics among the multiple sets of candidate movement characteristics corresponds to a route optimization score.

A method for determining a safe state for a vehicle includes disposing a camera at a vehicle and disposing an electronic control unit (ECU) at the vehicle. Image data is captured via the camera and provided to the ECU. An image processor of the ECU processes captured image data. A condition is determined via processing at the image processor of the ECU captured image data. The condition comprises a shadow present in the field of view of the camera within ten frames of captured image data or a damaged condition of the imager within two minutes of operation of the camera. The condition is indicative of a condition where processing of captured image data degrades in performance. The ECU determines a safe state for the vehicle responsive to determining the condition.

A method and a device are for identifying objects from camera images, e.g. for vehicle driver assistance systems. The method involves: capturing a series of camera images, determining a plurality of planes in space by associating adjacent corresponding features in at least two consecutive camera images with a given one of the planes, determining a relative translation vector of a plane, and identifying dynamic objects in the camera images based on the relative translation vector of an associated plane.

A system and method for correcting oversaturated pixels in far-infrared (FIR) images captured by a shutterless FIR camera, the method comprising: capturing thermal images by a FIR sensor in the shutterless FIR camera; processing the thermal images, by the shutterless FIR camera to determine pixel value and at least a shutterless sunburn correction, wherein the shutterless sunburn correction removes oversaturated pixels based on pixel-by-pixel analysis of the thermal image; and sending the processed thermal images to an output device.

An ECU is applied to a vehicle system that is provided with lateral sensors which acquire distance information expressing a distance to an object that is located at a position on a lateral side of a vehicle. When distance information on the object is acquired by the lateral sensors, a judgement is made by the ECU as to whether or not the object is a predetermined moving object that moves relative to the vehicle. The ECU determines that the object is a target to be subjected to contact avoidance processing for avoiding contact with the object, based on a result of judging whether or not the object for which the distance information is acquired is a moving object.

The invention relates to a stereo camera device for a motor vehicle, comprising: a first and a second mono camera, the first and second mono cameras each having at least one optical system and an image sensor for sensing images; an evaluation device, the image sensor of the first mono camera and the image sensor of the second mono camera being connected to the evaluation device in order to transmit image data of the sensed images. The evaluation device is designed: to evaluate a first stereo image from the image data transmitted to the first and second mono camera, using a stereo camera algorithm; to evaluate a second stereo image from the image data transmitted to the first mono camera, using a mono algorithm; to determine image depth information from each of the first and the second stereo images; and to check that the image depth information that has been determined from the first and the second stereo images corresponds.

A target detection device including a radar device and a monocular camera, including: a first detecting section detecting a position of a radar detection target; a second detecting section detecting a position of an image detection target which is a specific target; and a determination section that when the radar detection target and the image detection target are provisionally determined to be an identical target, and the image detection target is determined to be a predetermined type of target, determines that the radar detection target and the image detection target are not an identical target, and when the radar detection target and the image detection target are provisionally determined to be an identical target, and a predetermined target determination section determines that the image detection target is not the predetermined type of target, determines that the radar detection target and the image detection target are an identical target.