A method, apparatus, and system for estimating a moving speed of a detected pedestrian at an autonomous driving vehicle (ADV) is disclosed. A pedestrian is detected in a plurality of frames of point clouds generated by a LIDAR device installed at an autonomous driving vehicle (ADV). In each of at least two of the plurality of frames of point clouds, a minimum bounding box enclosing points corresponding to the pedestrian excluding points corresponding to limbs of the pedestrian is generated. A moving speed of the pedestrian is estimated based at least in part on the minimum bounding boxes across the at least two of the plurality of frames of point clouds. A trajectory for the ADV is planned based at least on the moving speed of the pedestrian. Thereafter, control signals are generated to drive the ADV based on the planned trajectory.

A method, apparatus, and system for estimating a moving speed of a detected pedestrian at an autonomous driving vehicle (ADV) is disclosed. A pedestrian is detected in a plurality of frames of point clouds generated by a LIDAR device installed at an autonomous driving vehicle (ADV). In each of at least two of the plurality of frames of point clouds, a minimum bounding box enclosing points corresponding to the pedestrian excluding points corresponding to limbs of the pedestrian is generated. A moving speed of the pedestrian is estimated based at least in part on the minimum bounding boxes across the at least two of the plurality of frames of point clouds. A trajectory for the ADV is planned based at least on the moving speed of the pedestrian. Thereafter, control signals are generated to drive the ADV based on the planned trajectory.

The technology disclosed relates to determining positional information of an object in a field of view. In particular, it relates to measuring, using a light sensitive sensor, one or more differences in an intensity of returning light that is (i) emitted from respective directionally oriented non-coplanar light sources of a plurality of directionally oriented light sources that have at least some overlapping fields of illumination and (ii) reflected from the target object as the target object moves through a region of space monitored by the light sensitive sensor, and recognizing signals in response to (i) positional information of the target object determined based on, a first position in space at a first time t0 and a second position in space at a second time t1 sensed using the measured one or more differences in the intensity of the returning light and (ii) a non-coplanar movement of the target object.

Techniques are described for time-of-fly sensors with hybrid refractive gradient-index optics. Some embodiments are for integration into portable electronic devices with cameras, such as smart phones. For example, a time-of-fly (TOF) imaging subsystem can receive optical information along an optical path at an imaging plane. A hybrid lens can be coupled with the TOF imaging subsystem and disposed in the optical path so that the imaging plane is substantially at a focal plane of the hybrid lens. The hybrid lens can include a less-than-quarter-pitch gradient index (GRIN) lens portion, and a refractive lens portion with a convex optical interface. The portions of the hybrid lens, together, produce a combined focal length that defines the focal plane. The hybrid lens is designed so that the combined focal length is less than a quarter-pitch focal length of the GRIN lens portion and has less spherical aberration than either lens portion.

A method for recognizing an object located in an object space includes emitting a distance measuring pulse into the object space by a signal time-of-flight based distance measuring unit. The object is provided with a marker which, in response to the influence of the distance measuring pulse, emits electromagnetic marker radiation in which object information for the object recognition is stored. The method further includes recording the marker radiation by an electrical radiation detector and the object information for object recognition being assigned to the object.

A laser measuring set for measuring a distance from an object includes a pulse laser for emitting a laser pulse at the beginning of a measuring cycle; an optical sensor having at least one detection unit for generating detection signals; a coincidence recognition stage for generating coincidence signals, wherein during the measuring cycle, one of the coincidence signals is generated each time the detection signals generated by the detection unit reach at least a preset coincidence depth within a coincidence time; a coincidence time presetting stage for presetting the coincidence time for the coincidence recognition stage, the coincidence time presetting stage being configured such that the coincidence time monotonically increases during the measuring cycle; and travel-time measuring set for determining the distance on the basis of a travel-time measurement of the coincidence signals.

Devices, methods, and computer program products for measuring the speed of an object. A speed measuring device includes a rangefinder module configured to measure distances from the device to a target object. Activating the device causes the device to measure a first distance from the device to the object along a first line-of-sight, and a second distance from the device to the object along a second line-of-sight. The device determines an angular displacement between the first line-of-sight and the second line-of-sight, and one or more of an elapsed time between measuring the first distance and measuring the second distance and a radial velocity of the object. The device then determines the absolute speed of the object based on the first distance, the second distance, the angular displacement, and one or more of the elapsed time and radial velocity.

In one embodiment, a method for solution inference using neural networks in LiDAR localization includes constructing a cost volume in a solution space for a predicted pose of an autonomous driving vehicle (ADV), the cost volume including a number of sub volumes, each sub volume representing a matching cost between a keypoint from an online point cloud and a corresponding keypoint on a pre-built point cloud map. The method further includes regularizing the cost volume using convention neural networks (CNNs) to refine the matching costs; and inferring, from the regularized cost volume, an optimal offset of the predicted pose. The optimal offset can be used to determine a location of the ADV.

In general, the present disclosure is directed to a time of flight (ToF) sensor arrangement that may be utilized by a robot (e.g., a robot vacuum) to identify and detect objects in a surrounding environment for mapping and localization purposes. In an embodiment, a robot is disclosed that includes a plurality of ToF sensors disposed about a housing of the robot. Two or more ToF sensors may be angled/aligned to establish overlapping field of views to form redundant detection regions around the robot. Objects that appear therein may then be detected by the robot and utilized to positively identify, e.g., with a high degree of confidence, the presence of the object. The identified objects may then be utilized as data points by the robot to build/update a map. The identified objects may also be utilized during pose routines that allow the robot to orient itself within the map.

A pulsed-light detection and ranging apparatus comprises an optical detector arranged to generate, when in use, time-series data in response to an optical pulse incident thereupon. A processing resource is also provided and arranged to support a pulse analyser (132). The pulse analyser (132) is arranged to identify (134) an inflection point of a pulse described by the time-series data. The pulse analyser (132) is further arranged to calculate (138) a distance based upon determined inflection point relative to a time axis associated with the time-series data.