Merging LIDAR information and camera information

Abstract

A system for an autonomous vehicle comprising merging LiDAR information and camera information. A vehicle including at least one LiDAR device configured to detect electromagnetic radiation; at least o

Description

[0114] As shown m FIG. 13A, vehicle 1302 includes LiDAR system 1304. In some embodiments, the LiDAR system 1304 of vehicle 1302 is the same as, or similar to, the LiDAR system 602 of AV 100. In some embodiments, one or more of the functions described with respect to the operation of vehicle 1302 are performed (e.g., partially, completely, etc.) by at least one processor of an image merging system 1350 of the vehicle 1302. In some embodiments, the at least one processor are the same as, or similar to, computing devices 146 of AV 100. Additionally, or alternatively, one or more of the functions described with respect to the operation of vehicle 1302 are performed (e.g., completely, partially, etc.) by a processor of a remote server. In some embodiments, the remote server is the same as, or similar to, the cloud server 136. In some embodiments, the remote server is the same as, or similar to, the cloud computing environment 200.

[0115] The at least one LiDAR system 1304 emits electromagnetic radiation in the form of light within a field of view 1306 of the LiDAR system 1304. The tight is then reflected off an object 1300 (e.g., a traffic light), and received by the LiDAR. system 1304. In this way, the LiDAR system 1304 detects electromagnetic radiation reflected from objects 1300 proximate (e.g., 1 meter, 2 meters, or another distance) to the vehicle 1302 and generates LiDAR information based on the detected electromagnetic radiation. The LiDAR system 1304 is in communication with the image merging system 1350 of the vehicle 1302 and LiDAR information is transmitted to the image merging system 1350 for processing (e.g., a signal is received by the at least one processor of the image merging system 1350 from the LiDAR system 1304 representing the LiDAR information). In this way, the image merging system 1350 analyzes the received LiDAR information to detect objects 1300 proximate to the vehicle 1302. In some cases, the LiDAR system 1304 generates LiDAR information associated with a point cloud based on LiDAR system 1304 receiving reflected light. In some cases, the LiDAR system 1304 transmits the LiDAR information associated with the point cloud to image merging system 1350 of the vehicle 1302. The image merging system 1350 generates a LiDAR image (e.g., as shown in FIG. 13B) based on the received LiDAR information. |0116] FIG. 13B is an example of a LiDAR image 1320 which represents (i) the heading (e.g., pitch and yaw of the LiDAR system 1304) of each LiDAR point of the LiDAR information relative to the vehicle 1302 in an 2D dimensional coordinate frame and (h ) the distance of each LiDAR point from the vehicle 1302 as a relative intensity of each LiDAR point. The relative intensity' is represented using a shading scale. For example, an object closest to the .LiDAR system 1304 (and. the vehicle 1302) is shown using a first shade 1308, an object furthest is shown using a second shade 1310, and objects between these two distances are shown using an interpolated shade. Portions of the LiDAR image 1312 with no shading 1312 did not receive any substantial reflections of light.

[0117] In some embodiments, LiDAR system 1304 determines that a reflection of a LiDAR point is substantial when a signal ratio of received light relative to transmitted light is greater than or equal to a threshold value. Additionally, or alternatively, in some embodiments, LiDAR. system 1304 determines that a reflection of a LiDAR point is not substantial when the signal ratio is less than or equal to the threshold value. In some examples, when the LiDAR system 1304 transmits light with unit power, and receives light less than one-one millionth of the unit power of the transmitted light, the received reflection is considered insubstantial (e.g., less than -60 dB using the formula I01ogio(received signal / transmitted signal)). The portions of the LiDAR image 1320 associated with no LIDAR information 1312 are below the threshold value

[0118] In some embodiments, the threshold value is varied to accommodate for a changing noise floor or the LiDAR system 1304 (e.g., electronic noise, analog to digital conversion, RF interference, etc.). The signal to noise ratio (SNR) is based on the noise floor. In some examples, electronic noise is so large (e.g., from a grounding problem) that the LiDAR system 1304 cannot distinguish received signals (and hence received LiDAR information) from the electronic noise. As a result, in this case, the threshold is increased above the electronic noise based on the electronic noise.

[0119] In some embodiments, the threshold value is varied to accommodate a changing environment of the LiDAR system 1304 (e.g., as the vehicle 1302 moves through the environment.) In some examples, the environment 1302 includes structures that do not reflect light well (e.g., non-r effective structures, no structures at all, etc.) and, as a result, substantial signals are not received by the LiDAR system 1304. In this case, the threshold is increased so that the structures are better resolved by the LiDAR system 1304. (0120] FIG. 13B shows two regions 1314, 1316 that represent traffic lights mounted to horizontal poles at an intersection. The traffic lights are resolved using a plurality of LiDAR points of the LiDAR information. In this scenario, a collection of one or more LiDAR points are represented as a LiDAR point cluster. In some embodiments, a LiDAR point cluster is determined by the image merging system 1350 based on a heading (e.g., pitch and yaw) relative to the vehicle 1302 of each respective LiDAR point and distance from the vehicle 1302 to each respective LiDAR point. In some examples, the image merging system 1350 determines that four LiDAR point clusters of objects (e.g., the two traffic lights in region 1314 and the two traffic lights in region 1316) are present in the LiDAR information.

[0121] The image merging system 1350 of the vehicle 1302 determines the heading and distance from the vehicle 1302 to each of the LiDAR point clusters of the LiDAR information. In some examples, an orientation is determined based on the heading and distance information. In some examples, the orientation represents a direction the object is facing. For example, traffic signals (e.g., red, yellow, green lights, etc.) of a traffic light are directed in the direction the traffic light is facing (e.g., towards oncoming vehicles). As another example, a stop sign includes a surface with a signal or message that is intended to be viewed in a direction normal to the surface (e.g., towards oncoming vehicles). In this scenario, the orientation represents a direction normal to the plane of the surface with the signal or message. In some examples, an average value of heading, distance, and orientation of each respective LiDAR point is used by the image merging system 1350 to determine the heading, distance, and orientation of the LiDAR point cluster. In some examples, this average value of heading and orientation includes a distance from the LiDAR system 1304 to one or more portions of a surface of an object.

[0122] In this way, the location (e.g., heading and distance) of LiDAR points indicate that one or more of a plurality of objects (e.g., a person, a trashcan, a sign, etc.) are present at a certain location within the environment of the vehicle 1302. For example, in the case of a traffic light 1300 as shown m FIG. 13 A, the image merging system 1350 of the vehicle 1302 processes the received LiDAR information to determine a plurality of distances between the vehicle 1302 and one or respective surfaces of the traffic lights.

[0123] FIG. 14A shows at least one camera system 1322 mounted on the vehicle 1302. In some embodiments, the camera system 1322 is the same as, or similar to, the camera system 122 of AV 100. The camera system 1322 is in communication with the image merging system 1350 the vehicle 1302 and transmits information as camera information to the at least one processor for processing (e.g., a signal is received by the at least one processor of the image merging system 1350 from the camera system 1322 representing the camera information). In this way, the image merging system 1350 analyzes the received camera information to detect objects 1300 proximate to the vehicle 1302 within the field of view 1324 (e.g., m the example shown, the camera system 1322 captures an image of the same object 1300 shown in FIG. 13A.) The image merging system 1350 generates a camera image (e.g., as shown in FIG. 14B) based on the received camera information.

[0124] FIG. 14B shows a typical output image 1330 of the camera system 1322 after a segmentation process has identified one or more traffic lights in the image. The traffic lights are associated with (e.g., represented by) the regions 1326, 1328 in the image 1330. For example, the image merging system 1350 of vehicle 1302 analyzes the image 1330 (e.g., using edge detection, color changes, contest gradients, etc.) to determine geometric features (e.g., surfaces, edges, vertices, etc.) and subsequently determine which objects 1300 are traffic lights based on knowledge of these geometric features (e.g., via an annotation process of the perception module).

[0125] An important aspect of this specification relates to merging, by the image merging system 1350, the LiDAR information from the LiDAR system 1304 and the camera information from the camera system 1322, This enables the image merging system 1350 to determine the orientation, heading, and / or position of objects of particular interest to the vehicle 1302 with increased accuracy compared to determining these same parameters using either the LiDAR information or the camera information alone. In this way, the image merging system 1350 associates geometric features represented in the image information (e.g., FIG. 14B) with the LiDAR points and / or LiDAR point clusters from the LiDAR information (e.g., FIG. 13B).

[0126] In an embodiment, the image merging system 1350 of the vehicle 1302 identifies the detected object 1300 based on the LiDAR information and camera information. In some examples, the identification associates the detected object 1300 with a traffic light.

[0127] In an embodiment, the LiDAR information, the camera information, and / or the merged image is used by the image merging system 1350 of the vehicle 1302 to determine classification information of the object. The classification information is representative of a level of interest of the object within the LiDAR information and camera information as previously described. In some examples, the annotation and segmentation process determines the classification information and transmits the classification information to the image merging system 1350 for processing.

[0128] In some examples, the classification is indicative of the object moving in the environment (e.g., vehicles, swaying trees, pedestrians, etc.) or stationary in the environment (e.g., traffic lights, traffic signs [stop signs, yield signs, speed limit signs], cross walks, etc.). In an embodiment, moving objects are detected by the image merging system 1350 of the vehicle 1302 based on acquired camera information and acquired LiDAR information over a time period (e.g., camera information and LiDAR information is acquired multiple times to determine a change in position of the object over time. In this scenario, in some examples, the image merging system 1350 determines that an object moved 0.1 m towards the vehicle '1302 in 2 seconds. In some examples, moving objects are detected based on tracking an identification of the annotation and classification process (e.g., the annotation process determines that an object is a vehicle using the features of the object such as license plates, edges of the object, head lights, etc.).

[0129] In an embodiment, the classification process groups the object into one or more categories (e.g., category 1 are traffic lights, category 2 are traffic signs, and category 3 are everything else). For example, if the vehicle 1302 becomes aware of several objects ahead in category' Lit triggers a process to determine the color and position of the light emitted from the traffic light to determine whether the traffic light is signaling the vehicle 1302 to go, yield, or stop. In an embodiment, directional indicators of the traffic light (e.g., a green left arrow) are detectable by analyzing the features of the light emitted by the traffic light. In the scenario where the detected object is a traffic light, in some examples, the image merging system 1350 determines that a change to the vehicles trajectory is not needed (e.g., no need to slow down yet) if the traffic light is yellow and is 500 m away from the vehicle 1302. On the other hand, in some examples, the image merging system 1350 instructs the vehicle to perform an update to the trajectory in order to enable the vehicle 1302 to immediately stop if a traffic light is red and is 20 m away from the vehicle 1302.

[0130] In some examples, the image merging system 1350 of the vehicle determines an available response time based on a distance from the vehicle 1302 to the object (e.g., as determined by the image merging system 1350 using the LiDAR information) and a current velocity of the vehicle 1302. In this way, the available response time represents how much time is available to react to the detected object. In some examples, the trajectory of the vehicle 1302 is updated based on the available response time (e.g., if there is not enough time to stop in the left turning lane, then switch lanes and proceed straight through the intersection).

[0131] In an embodiment, if the vehicle 1302 becomes aware of several objects ahead in category 2, the image merging system 1350 of the vehicle 1302 determines (or causes a determination of) what the traffic sign says or means. In this scenario, the image merging system 1350 interprets the words and / or the expressions from the sign. In some examples, the traffic sign is a pedestrian cross walk sign and is detected by extracting features of the sign (e.g., the shape of the sign) and performing image processing via the annotation process (e.g., the received camera information, LiDAR information, and / or merged information is compared with known images of cross-walk indicators).

[0132] In an embodiment, the image merging system 1350 of the vehicle 1302 establishes an alert level based on the classification of objects. In some examples, the alert level is based on received annotation information. In some examples, the image merging system 1350 of the vehicle 1302 places the vehicle 1302 on high alert (e.g., higher than normal) for pedestrians in the area. In some examples, the image merging system 1350 controls the vehicle 1302 to slow down in response to being on high alert. In some examples, the image merging system 1350 transmits these control signals to a separate vehicle control for controlling the vehicle 1302.

[0133] In an embodiment, the image merging system 1350 of the vehicle 1302 establishes a yield level representing whether or not the vehicle 1302 has the right of way. In some examples, if the image merging system 1350 determines that an upcoming traffic sign is a yield sign, the image merging system 1350 establishes tire yield level and controls the vehicle 1302 to yield to other vehicle. In some examples, the image merging sy stem 1350 transmits the control signal to a separate vehicle control for controlling the vehicle 1302.

[0134] In some examples, traffic signs have instructions associated to them (e.g., school zone sign causes the vehicle 1302 to slow down, stop sign causes the vehicle 1302 to stop, etc.) and the instructions are received by the image merging system 1350 of the vehicle 1302, interpreted, and used for controlling operations of the vehicle 1302. In some examples, the instructions of traffic signs are performed by the image merging system 1350 via a look-up table or database. In some examples, the instructions are retrieved by the image merging system 1350 from a map.

[0135] In an embodiment, a temporary distinction is included in the classification information. The temporary distinction is indicative of a temporary object. In some examples, the image merging system 1350 uses the temporary distinction to control the vehicle cautiously. For example, in the case of a construction zone, the vehicle slows down. In some examples, the temporary distinction is based on the object classification (e.g., only apply a temporary distinction if the object is classified using category 1 and / or category’ 2).

[0136] In some examples, the image merging system 1350 of the vehicle 1302 determines that a traffic sign in category 2 is temporary based on movement of the traffic sign (e.g., such as when worker waves the sign or flips the sign, the sign is attached to a moving construction vehicle, etc. ). In some examples, the image merging system 1350 infers that the traffic sign should not move (e.g., by determining that the traffic sign is rigidly connected to the ground based on a presence of a rigid connection (a pole) between the traffic sign and the ground) and a. determination of movement results in a temporary distinction. In some examples, the image merging system 1350 infers that the traffic sign should not move based on receiving an indication that the traffic sign is a permanent structure from the server, database, map, or from the traffic sign itself (e.g., via a broadcasting wireless signal from the traffic sign).

[0137] In some embodiments, the image merging system 1350 of the vehicle 1302 accounts for a timing difference between the acquisition of the camera information by the camera system 1322 and the LiDAR. information by the LiDAR system 1322, In some examples, the image merging system 1350 determines the merging process of the LiDAR information and the camera information based on the timing difference. Aspects related to the timing difference are further explained with respect to FIGS. 15A-15D below.

[0138] FIGS. 15A-15D show a timing sequence of steps that occur between acquiring the LiDAR information from the LiDAR system 1304 and acquiring the camera information from the camera 1322.

[0139] FIG. ISA shows the traffic light 1300 within a field of view 1402 of the LIDAR system 1304 of the vehicle 1302. In some examples, the field of view 1402 represents an outline of which objects are detectable by the Li DAR system 1304. In an embodiment, the LiDAR. system 1304 is pitched downward to emphasize objects on the ground of the environment over objects in the air (e.g., above the ground). The LiDAR system 1304 acquires LiDAR information of the traffic light 1300 and the LiDAR information is transmitted to the image merging system 1350 of the vehicle 1302.

[0140] FIG. 15B shows a field of view 1404 of the camera system 1322 of the vehicle 1302. The field of view 1402 is typically different from the field of view 1404 because the LiDAR system 1304 and camera system 1322 typically have different optical configurations. At a particular point in time, the processor of the vehicle 1302 instructs the LiDAR system 1304 and the camera system 1322 to acquire information about the environment of the vehicle 1302. As shown in FIGS. 15A and 15B, if the LiDAR system 1304 and the camera system 1322 acquire information at the same time, a scenario could exist where the field of view '1402 of the LiDAR system 1304 includes the traffic light 1300, but the field of view 1404 of the camera system 1322 does not include the traffic light 1300. In this context, field of view 1404 denotes regions where the detectable objects are within focus and resolvable with a threshold number of pixels (e g., at least 100 pixels).

[0141] FIG. 15C shows a situation where the vehicle 1302 has moved closer to the traffic light 1300 than shown in FIGS. 15A and 15B. In this scenario, the LiDAR system 1304 is unable to completely resolve the traffic light 1300 because at least a portion of the traffic light 1300 is outside of the field of view 1402 of the LiD.AR system 1304.

[0142] FIG. 15D show's a composite where the image merging system 1350 of the vehicle 1302 instructs the LiDAR system 1304 to acquire LiDAR information first (e.g., before the camera system 1322). This is represented in state A of FIG. 15D. Then, the image merging system 1350 of the vehicle 1302 instructs the camera system 1322 to acquire camera information after a time has passed. This is shown in state B of FIG. 15D. In this way, the acquisition of LiDAR information from the LiDAR system 1304 and camera information from the camera 1322 is temporally offset to account for the field of view differences between the LiDAR system 1304 and the camera system 1322. In the scenario shown in FIG. 15D, LiDAR information is acquired from the LiDAR system 1304 at state A and, about 2 seconds later, camera information is acquired from the camera system 1322 at state B.

[0143] In some examples, the image merging system 1350 of the vehicle 1302 determines from the LiDAR information that a traffic light is 10 meters away but camera information received from the camera system 1322 is insufficient to resolve the traffic light. In this scenario, perhaps only a few pixels of the camera information represent the traffic light. In some examples, less than 10% of pixels of the camera information represent features of the traffic light and is considered insufficient to resolve the traffic light.

[0144] In an embodiment, the LiDAR information includes information from more than one scan (e.g., 2 to 20 LiDAR scans such as 10 LiDAR scans). In an embodiment, the time delay between the acquisition of the LiDAR information and the camera information is based on the travel time of the vehicle 1302. In some examples, the travel time is estimated, by the image merging system 1350, based on a location difference between a location of the vehicle 1302 when acquiring the LiDAR information and a location of the vehicle 1302 when acquiring the camera information. In some examples, the travel time is estimated, by the image merging system 1350, based on. a velocity of the vehicle 1302.

[0145] In some embodiments, the image merging system 1350 of the vehicle 1302 determines an accuracy based on the number of pixels representing features of the traffic light 1300. In some scenarios, the image merging system 1350 of the vehicle 1302 determines the accuracy based on the number of pixels representing features of the traffic light 1300 exceeding a threshold value (e.g., 100 pixels is associated with good accuracy, 10 pixels is associated with poor accuracy). In these examples, the accuracy is based on a number of pixels associated with the object 1300 (e.g., used to resolve the object).

[0146] In some examples, the accuracy is based on the distance between the vehicle 1302 and the traffic light 1300 (e.g., the image merging system 1350 of the vehicle 1302 determines an accuracy based on a physical distance to the traffic light 1300). In some examples, as the vehicle 1302 moves closer to the traffic light 1300 (e.g., after 2 seconds has elapsed), the camera information is reacquired and more (e.g., several hundred) pixels are associated with the traffic light 1300 compared to the image information from 2 seconds prior.

[0147] In some examples, the image merging system 1350 determines a confidence based on the accuracy and / or information from an external server, database, or map. For examples, m some cases, when the image merging system 1350 receives an indication from the server, database, or map that a traffic light is approaching, the image merging system 1350 determines if the accuracy of the detected object is above a threshold value (e.g., based on a number of pixels, etc.). In the case of the accuracy of the detected object being above the threshold value, a high confidence is associated with the detected object. On the other hand, in the case of the accuracy of the detected object being below the threshold value, a low confidence is associated with the detected object, in these cases, the confidence represents how confident the vehicle 1302 is that the object 1300 ahead is in fact a traffic light. In some examples, the trajectory of the vehicle 1302 and / or the path planning of the vehicle 1302 is based on the confidence. In some examples, the vehicle 1302 ignores detected objects when the confidence is low (e.g., below 5%). In some examples, the vehicle 1302 uses additional resources (passenger confirmation, external map information, etc.) to confirm the presence of the object when the confidence is low'.

[0148] In some embodiments, the image merging system 1350 of the vehicle controls a zoom feature (e.g., optical zoom or digital zoom) of the camera system 1322 to zoom-m on the object 1300. In some cases, controlling the zoom feature is based the accuracy and / or the confidence. For example, when the accuracy (or confidence, or both accuracy and confidence) is / are below a threshold, the image merging system 1350 determines that a second look is warranted and controls (or instructs) the zoom feature of the camera system 1322 (or a second camera system) to optically zoom in with a 2X magnification. In some examples, the image merging system 1350 of the vehicle controls (or instructs) the camera system 1322 to pan to an area of the detected object based on the camera information (or merged information). Once the camera system 1322 is ready (zoomed and / or panned), the camera system 1322 then generates second camera information and transmits the second camera information to the image merging system 1350 of the vehicle. The image merging system 1350 analyzes the second camera information similar to the first camera information. In some embodiments, a second camera system is used to acquire the second camera information.

[0149] Similarly, in some embodiments, a second LiDAR system is used to acquire second LiDAR information. For example, when the accuracy (or confidence, or both accuracy and confidence) is / are below a threshold, the image merging system 1350 determines that a second look is warranted and controls (or instructs) a zoom feature of the LiDAR system 1304 (or a second LiDAR system) to optically zoom in with a 2X magnification. In some examples, the image merging system 1350 of the vehicle controls (or instructs) the LiDAR system 1304 to pan to an area of the detected object based on the LiDAR information (or merged information). Once the LiDAR system 1304 is ready (zoomed and / or panned), tiie LiDAR system 1304 then generates second LiDAR information and transmits the second LiDAR information to the image merging system 1350 of the vehicle. The image merging system 1350 analyzes the second LiDAR information similar to the first LiDAR information. 10150] In an embodiment, LiDAR information is continuously acquired for a period of time (e.g., up to 10 seconds) and stored within a memory buffer of the vehicle 1302. In some embodiments, LiDAR information is retrieved, by the image merging system 1350 of the vehicle 1302, from the memory buffer for processing.

[0151] Once the image merging system 1350 of the vehicle 1302 has access to LiDAR information and camera information representing the environment of the vehicle 1302 (e.g., by acquisition or by buffer), the image merging system 1350 begins a merging process. The image merging system 1350 associates at least one portion of the LiDAR information with at least one pixel of the camera information to improve the vehicle’s understanding of the environment. Typically, the at least one pixel of the camera information represents an object of particular interest (e.g., a stop sign, a traffic light, a vehicle, etc.).

[0152] Referring back to the example shown in FIG. 15D, in this scenario, the camera information associated with the camera system 1322 is merged, by the image merging system 1350 of the vehicle 1302, with the corresponding LiDAR information from 2 seconds prior. In other examples, the acquisition of the LiDAR information and the camera information occur simultaneously

[0153] Through the merging process, the image merging system 1350 synchronizes the respective fields of view' of the LiDAR system 1304 and the camera system 1322 with each other. In some examples, the image merging system 1350 compares image features (e.g., edges, faces, colors, etc.) of the camera information with the LiDAR information previously acquired to determine the synchronization.

[0154] FIG. 16A is an illustration of the merging process of LiDAR information from the LiDAR system 1304 with camera information from the camera system 1322. As previously described, the image merging system 1350 associates at least one portion of the LiDAR information with at least one pixel of the camera information. In some examples, the image merging system 1350 of the vehicle 1302 merges the LiDAR information with the camera information based on a relative distance between the LiDAR system 1304 and the camera system 1322. In some examples, the image merging system 1350 of the vehicle 1302 merges the LiDAR information with the camera information based on a ratio of the field of view of the LiDAR system 1304 to the camera system 1322.

[0155] In an embodiment, the image merging system 1350 determines an alignment between respective features of the LiDAR information and the camera information based on features represented in both the LiDAR information and the camera information. In some examples, feature gradients such as sharp edges are distinguishable in both the LiDAR information and the camera information and are used by the image merging system 1350 to merge the LiDAR information to the camera information. In this scenario, edges in the camera information are aligned with gradients in the LiDAR information. (0156] In some embodiments, the image merging system 1350 merges ail the LiDAR information with the camera information. In some embodiments, the image merging system 1350 merges portions of the LiDAR information with the camera information. For example, the image merging system 1350 selects regions of particular interest from the LiDAR information, merges these regions with the camera information, and discards the remaining portions of the LiDAR information. As described with respect to FIG. 13B above, in some examples, the regions of LiDAR information of particular interest are LiDAR point clusters. One or more LiDAR point clusters are present within the LiDAR information and are determined by the image merging system 1350 of the vehicle 1302,

[0157] In some embodiments, the image merging system 1350 determines a mapping for each LiDAR point cluster within the LiDAR information. In some examples, the merging process is performed independently for each LiDAR point cluster. In this scenario, the image merging system 1350 determines a mapping representing a best fit mapping between the respective LiDAR point cluster and the one or more pixels of the camera information. In this way, the image merging system 1350 associates object information of one or more pixels of the camera information with LiDAR point clusters of the LiDAR information. The merging of LiDAR information with camera information enables the image merging system 1350 to query a region of the merged information and retrieve respective LiDAR information and camera information. As an example, when the image merging system 1350 inquires about a status or property (e.g., color, intensity, position, etc.) of the regions within the merged LiDAR information and camera information, the image merging system 1350 receives information associated with both LiDAR information (such as distance information, time of LiDAR acquisition, etc.) associated with LiDAR points within the regions and camera information (such as color, intensity, 2D position, etc.). For example, in scenarios where the detected object is a traffic light, the image merging system 1350 inquires on the color of pixels associated with the merged information to infer which traffic instruction (e.g., green-go, yellow-yield, and red-stop) is being instructed by the traffic light. In these cases, color and intensity information is retrieved from the camera information part of the merged information and orientation and distance information is retrieved from the LiDAR information part of the merged information. In some examples, when the image merging system 1350 determines that the intensity is above a threshold, the image merging system 1350 controls the vehicle to respond to the traffic instruction.

[0158] FIG. 16B shows a composite image in which LiDAR information (represented using dots in FIG. 16B) from the LiDAR system 1304 is merged with camera information (represented using lines in FIG. I6B) from the camera system 1322, In particular, FIG. 16B shows that physical objects in the environment (e.g,, edges of buildings, traffic lights, etc.) are associated with both camera information and LiDAR information.

[0159] In the example shown in FIG. 16B, a pair of traffic lights 1504 within a region 1502 are associated with both camera information and Li DAR information. Bounding boxes, or edges, of the traffic lights 1504 are shown in a bolded outline. In some examples, the image merging system 1350 determines or receives bounding box information from the segmentation and annotation process. In some examples, the segmentation and annotation process of the vehicle 1302 identifies objects of interest (e.g., traffic lights 1502, stop signs, pedestrians, etc.) represented in the camera, information and associates one or more pixels of the camera information that represent these objects with bounding boxes. In some examples, the bounding box is determined based on the camera information without considering the LiDAR information. In some examples, the bounding box is determined based on objects of interest (e.g., a classification of an object of interest). For example, in some cases, the segmentation and annotation process determines bounding boxes for all detected objects, but the image merging system 1350 filters (ignores) all detected objects within certain categories (e.g., ignores all bounding boxes associated with category 3, etc.).

[0160] In an embodiment, the image merging system 1350 determines LiDAR information located within the bounding box (e.g., within the boundaries defined by the edges). In some examples, the image merging system 1350 determines a distance from the vehicle 1302 to the traffic lights 1502 based on an average distance associated with LiDAR information within the bounding box.

[0161] In an embodiment, the image merging system 1350 determines a distance in the environment from the vehicle 1302 to the object using the merged LiDAR information encapsulated by the bounding boxes. In some examples, the distance is used by the image merging system 1350 of the vehicle 1302 for control purposes and map annotation. In some examples, the image merging system 1350 controls the vehicle 1302 m response to the determined distance using the merged LiDAR information and camera information.

[0162] In some examples, the position of the object within a map is annotated, by associating each traffic light with a position in the environment, to denote that a traffic light exists at a certain position. In some examples, the map is a global map. In an embodiment, the map is used by other vehicles to anticipate the position of traffic lights within the environment. In some examples, the position of the object is determined by the image merging system 1350 of the vehicle 1302 based on the location of the vehicle 1302 (e.g., as received by a GPS sensor, as determined by a localization system or module, and / or the like), the position, orientation, and field of view of the camera system 1322, and the position, orientation, and field of view of LiDAR system 1304.

[0163] In an embodiment, the image merging system 1350 determines an instance identifier (id) of the object. The instance identifier of the object is used for tracking the object through the environment. In some examples, the instance identifier is a unique number for tracking purposes.

[0164] For example, when the vehicle 1302 identifies a new (e.g., not previously observed) traffic light, the image merging system 1350 of the vehicle 1302 assigns the new traffic light to category 1 along with the position information of the new traffic light. As the vehicle 1302 moves though the environment, the image merging system 1350 of the vehicle continuously observes the object and verifies which object is the same and which objects are new based on the instance identifier. In a scenario, the vehicle 1302 turns a corner and observes a new traffic light ahead. Subsequently, the image merging system 1350 determines this and assigns it to a new instance identifier (e.g., a numeral associated with the instance identifier is incremented).

[0165] In an embodiment, the instance identifier is based on the confidence. For example, as the vehicle 1302 gets closer to the object, the image merging system 1350 determines with a higher confidence that the object is actually a traffic light (e.g., based on a determination that there is more LiDAR information and camera information of the object available via the sensors of vehicle 1302).

[0166] In an embodiment, the image merging system 1350 continuously updates the position information of the object. For example, the estimated position associated with an object (e.g., the position of the object relative to a pre-generated 3D map, the position of the object relative to one or more buildings in an area, and / or the like) 10 meters away from the vehicle 1302 will be typically less accurate that the estimated position information when only 1 meter away. In this way, the image merging system 1350 retains position information with the highest confidence. In some examples, this position information is stored within memory of the vehicle 1302, or a remote database or map.

[0167] In an embodiment, the image merging system 1350 updates the classification over time. In some examples, the image merging system 1350 updates the classification of an object (e.g., from category 3 to category 2). For example, in a scenario, from a distant perspective the image merging system 1350 of the vehicle 1302 determines that a pedestrian is ahead and assigns the associated merged LiDAR information to category 3, but as the vehicle 1302 approaches the image merging system 1350 realizes that the pedestrian classification is incorrect and that the object is actually a stop sign and updates the classification information to category 2.

[0168] In an embodiment, the image merging system 1350 updates the instance identifier based on subsequent merged LiDAR information and camera information. In some examples, the image merging system 1350 of the vehicle 1302 determines that a single traffic light is ahead, but as the vehicle 1302 approaches, the image merging system 1350 realizes the traffic light is actually two traffic lights. In this scenario, the image merging system 1350 continuously determines whether the object has split into multiple objects. Here, the traffic light that was originally assigned to a single instance identifier by the image merging system 1350, is split into two traffic lights and each traffic light is assigned a unique instance identifier by the image merging system 1350 of the vehicle 1302. In this way, instance identifiers of the object are updatable.

[0169] In an embodiment, the image merging system 1350 determines an orientation of the object (e.g., an orientation representing a direction an instruction of the object is facing). In some examples, a traffic light ahead that is aimed at a lane other than the lane of the vehicle 1302 (e.g., a traffic instruction of the traffic light is not directed to the vehicle 1302) is used for updating a map despite not being aimed in the path of the vehicle 1302. In other words, in some cases, the presence and orientation of the traffic light is updated in the map regardless of whether the traffic light is in the path of the vehicle 1302.

[0170] In an embodiment, determining the orientation is based on features of the object itself (e.g., edges, colors, proportions, etc.). In an embodiment, the image merging system 1350 determines a direction based to the orientation. In some examples, an edge of a stop sign surface is used to infer a direction normal to the surface of the stop sign (e.g., inferred by computing a vector cross product of a surface spanned by the LiDAR point clusters representing the surface). In another example, an edge of a traffic fight is used to infer a direction of the traffic light (e.g., to represent where the traffic signal and hence traffic instruction is directed) In some examples, the image merging system 1350 controls the vehicle based on the orientation of the detected object. For example, if the image merging system 1350 determines that the direction the traffic light is directed toward the vehicle (e.g., by determining when a vector dot product between the direction the detected object, is directed and a direction of travel of the vehicle is above a threshold), the vehicle is controlled to respond to the traffic signal of the traffic instruction of the traffic signal. In other cases, the controller of the vehicle does not respond to the traffic instruction of the traffic light when the image merging system 1350 determines that the direction the traffic light is not directed toward the vehicle (e.g., by determining when a vector dot product between the direction the detected object is directed and a direction of travel of the vehicle is below a threshold).

[0171] In an embodiment, the image merging system 1350 of the vehicle 1302 builds a 3D representation of the object as the vehicle 1302 traverses the environment of the object. In an embodiment, the 3D representation is used to determine the orientation of the object.

[0172] FIG. 16C shows LiDAR information that is encapsulated by bounding boxes (not shown). In contrast with the LiDAR image 1320 shown in FIG. 13B that shows all LiDAR information, FIG. 16C shows LiDAR information that is encapsulated by the one or more bounding boxes. In this way, the one or more bounding boxes serve as a filter of LiDAR information and only LiDAR information within the one or more bounding boxes persist after this filtering process.

[0173] FIG. 16D shows a detail view of LiDAR information associated with one of the traffic lights of FIG. 16C. In particular, FIG. 16D shows each LiDAR point associated with a LiDAR point cluster representing a traffic light and respective properties for each LiDAR point.

[0174] The image merging system 1350 of the vehicle 1302 determines the respective properties based on the merged LiDAR information and camera information. In the example shown. each LiDAR point is annotated by the image merging system 1350 to include position information (e.g., x, y, z), classification information (e.g., a “class” parameter), and instance information (e.g., an “instance id” parameter). In some examples, each LiDAR point is annotated with other information (e.g., physical properties (e.g., color, size, etc.), status (e.g., temporary, last observed, etc.), etc.).

[0175] FIGS. 17A-17F illustrate an embodiment where the merging process is performed using LiDAR information and camera information representing a 360 degree view around the vehicle 1302. The image merging system 1350 of the vehicle 1302 forms a composite representation 1600 representing the merging of LiDAR information with camera information. In this example, the camera information represents six camera images acquired from the camera system 1322 of the vehicle 1302. In this example, FIG. 17B represents the environment directly m front of the vehicle 1302, FIG. 17E represents the environment directly behind the vehicle 1302, FIG. 17 A represents the environment to the front left of the vehicle 1302, FIG. 17C represents the environment to the front, right of the vehicle 1302, FIG. 17D represents the environment to the rear right of the vehicle 1302, and FIG. 17F represents the environment to the rear left of the vehicle 1302. The composite representation 1600 defines a complete 360 degree view around the vehicle 1302. The image merging system 1350 of the vehicle merges the LiDAR information with the camera information spanning the complete 360 degree view. In this way, the image merging system 1350 of the vehicle 1302 determines a distance of each respective object around a 360 degree view of the vehicle 1302.

[0176] FIG. 18 is a flowchart of an embodiment of an image merging process 1700 of merging LiDAR information with camera information. The image merging process 1700 is implementable using the at least one processor of the image merging system 1350 of the vehicle 1302 or the at least one processor of the remote server (e.g., cloud server 136 or cloud computing environment 200). While the image merging process 1700 illustrates a particular flow of information, embodiments of the merging L iDAR information with camera information are not limited to a particular flow of information.

[0177] The at least one camera 1322 is used to acquire camera information 1702. In some examples, the camera information 1702 represents a 360 degree view around the vehicle 1302.

[0178] The at least one LiDAR system 1304 is used to acquire LiDAR information 1704. In some examples, the image merging system 1350 of the vehicle receives vehicle position and orientation information 1712 of the vehicle 1302 along with the LiDAR information 1704.

[0179] The LiDAR information from the LiDAR system 1304 and the at least one camera 1322 are used in the merging process 1706. In other words, the LiDAR information from the LiDAR system 1304 and the at least one camera 1322 are received by the image merging system 1350 when the image merging system 1350 implements the process of the merging process 1706.

[0180] While performing the merging process 1706, the image merging system 1350 determines the best fit of merging the LiDAR information with the camera information and generates merged information based on the LiDAR information and the camera information.

[0181] The process 1700 includes a segmentation, annotation, and classification process 1708. In the example shown, the merged information from the merging process 1706 is used by the segmentation, annotation, and classification process 1708. However, in some examples, the segmentation, annotation, and classification process 1708 is performed before the merging process 1706 (e.g., using either camera information or LiDAR information alone).

[0182] In the example shown, the image merging system 1350 associates objects with three categories (category 1 being traffic lights, category 2 being traffic signs, and category 3 being everything else) while implementing the segmentation, annotation, and classification process 1708. The category information is included in classification information that is associated with each object. In this way, each object includes classification information. In some examples, an instance identifier is assigned as part of the segmentation, annotation, and classification process 1708 and included in the classification information.

[0183] In an embodiment, an object query process 1708 is included in the process 1700. While performing the object query process 1708, the image merging system 1350 iterates (e.g., loops) over each instance identifier in each category to determine an action of the vehicle 1302. In some examples, the image merging system 1350 updates a map 1710 with the object information (e.g., category, instance identifier, position and orientation). In some examples, this map is queried by other vehicles and respective processors of the other vehicles use this map to determine when to anticipate upcoming traffic lights or traffic signs. [01841 In an embodiment, the image merging system 1350 also determines whether to store the information associated with the object in a server (such as server 1714) or database (such as database 1716). In an embodiment, the image merging system 1350 stores the information associated with the object in the server or database. Similarly, in an embodiment, the image merging system 1350 retrieves object information from the server, database, or map, updates the object information accordingly, and sends the object information back to the server, database, or map.

[0185] FIGS. 19A-19B show' details of the global map 1710. The map 1710 includes roads of the environment. The map 1710 is shown with X coordinates on the horizontal axis and Y coordinates on the vertical axis. Z coordinates are not shown for brevity, but in. some embodiments, are included in the information stored in the global map 1710 to denote height of the object off’of the ground (or above sea level). In the example shown tn FIG, 19B, the global map 1710 includes the location of five objects within a region 1802. The live objects are stored in the global map 1710. When the image merging system 1350 retrieves object information from the map 1710, information regarding the five objects is transmitted to the image merging system 1350

[0186] In an embodiment, after the image merging system 1350 of the vehicle 1302 identifies a traffic light ahead, the image merging system 1350 compares the position of the traffic light with known instances of objects in the global map 1710. For example, if the global map 1710 indicates a traffic light at a particular location ahead, the image merging system 1350 of the vehicle 1302 adopts the instance identifier of the traffic light and downloads the associated information of the traffic light from the global map 1710. For example, in some cases the position information of an object included in the map is more accurate than the current position estimated by the image merging system 1350 of the vehicle 1302. In other examples, the object is too far away from the vehicle 1302 to determine an orientation of the object so it is downloaded by the global map 1710 instead and adopted by the image merging system 1350 as the orientation of the object. In this way, the global map 1710 is annotated with current, up to date, estimates of the position and orientation of all objects in category 1 and 2 that have been observed by an image merging system 1350 of a vehicle at least once.

[0187] In an embodiment, the number of times the object has been observed is also stored on the global map 1710. In some examples, the number of times the object has been observed is indicative of a probability that the object exists. For example, in some cases, if an object is observed at least once per day, the object is associated with a high probability that the object exists. On the other hand, in some cases, if an object is observed less than once per year (e.g., once a week, once a month, and / or the like), the object is associated with a low probability’ that the object exists. In some examples, if an object is removed, or is moved, the map is updated accordingly. In an embodiment, the object remains on the global map 1710, but the object is associated with information that indicates that the object may not exist since at least one instance exists where an image merging system 1350 of a vehicle did not detect the object (e.g., when there is contradicting information associated with an object). In an embodiment, dates and times associated with the information are stored on the global map 1710. In some examples, dates and time information are used to give the image merging system 1350 an indication of information reliability. In some examples, a newly identified object (e.g., observed two days ago) is considered more reliable that an object last observed a year ago.

[0188] In an embodiment, this information is part of the global map 1710 and used in the route planning process. In some examples, if the route planner receives an indication that ten traffic lights exist along a path, the route planner reroutes the vehicle 1302 for fuel and / or energy efficiency reasons and / or for travel time considerations. In an embodiment, passenger comfort is considered based on the number of objects identified in the global map 1710 along particular routes. In some examples, a route with ten stop signs is less comfortable to a passenger that has motion sickness than a route that is longer (distance-wise and / or time-wise) but includes less stops (e.g., highway travel).

[0189] FIG. 20 is a flowchart of the image merging process 1900 of LiDAR information with camera information.

[0190] The image merging process 1900 is performed by the at least one processor of the image merging system of a vehicle (such as the image merging system 1350 of vehicle 1302) having at least one LiDAR device (such as the LiDAR device or LiDAR system 1304) configured to detect electromagnetic radiation (e.g., light) in the ultraviolet, infrared, or laser spectra, or any other kind of electromagnetic radiation. The vehicle includes at least one camera (such as the camera 1322) configured to generate camera information of the objects proximate to the vehicle in a field of view of the camera. The vehicle includes at least one processor configured to implement the operation / steps of the image merging process 1900. The at least one processor performs some or all of the operations / steps of the image merging system. 10191] The image merging system receives 1902 LiDAR information from the at least one LiDAR device. In some examples, the LiDAR information covers a 360 degree azimuth around the vehicle. In an embodiment, the LiDAR information includes multiple LiDAR scans using the LiDAR system 1304 and is received in real-time by the image merging system.

[0192] The image merging system receives 1904 object information of an object proximate to the vehicle based on camera information from the at least one camera. In an embodiment, the object information includes at least one pixel of the camera information representing the object. Tn an embodiment, the object information includes categorization information representing a classification of the object. In an embodiment, the image merging system receives the camera information.

[0193] In some examples, categorization information associates a pixel with an object. For example, if at least one pixel of the image is determined to represent a traffic light, this association is included in the categorization information. In an embodiment, the segmentation and annotation process determines which pixels of the image correspond to objects (e.g. a first sei of pixels correspond to a traffic light, a second set of pixels correspond to a pedestrian, etc.).

[0194] In an embodiment, the classification represents a category for these objects (e.g., category' 1 is a traffic tight, category 2 is a traffic sign, and category 3 is everything else). In a broader context, a category is grouping of objects with at least one common feature. For example, in an embodiment, categories 1 and 2 include all traffic instructions (e.g., stop sign, yield sign, school zone sign, traffic light, speed limit signs, merge lane signs, etc.) and category 3 includes everything that is not a traffic instruction (e.g., a tree, a building, a vehicle, a parking meter, etc.). In this context, the common feature is traffic instruction. In general, in some examples, a traffic instruction refers to any explicit instruction received by the vehicle used to navigate through the environment (e.g., a stop sign includes an explicit traffic instruction to stop, a. yield sign includes an explicit instruction to yield to other vehicles, a speed limit sign includes an explicit instruction not to exceed the stated speed limit, and / or the like).

[0195] In an embodiment, the image merging system determines 1906 if the categorization information of the object is associated with a traffic instruction. In some examples, the operation of the vehicle is based on the whether or not a traffic instruction is ahead of the vehicle (e.g., stop at a stop light, slow down in a school zone, etc.). In some examples, the traffic instruction is a traffic light or a traffic sign. In an embodiment, the image merging system determines a traffic signal of the traffic instruction based on the merged information. In some examples, the image merging system infers the signal of the traffic light based on a color of at least one pixel in the merged information (e.g., red, yellow, or green means stop, yield, and go, respectively). In an embodiment, the image merging system infers the traffic signal based on a position of peak light intensity of light emitted from the object (e.g., if a bright light is emitted near the top of the object, it is inferred to represent a stop instruction).

[0196] In an embodiment, the control circuit of the vehicle is further configured to operate the vehicle based on the traffic signal of the traffic instructions.

[0197] In accordance with determining that the categorization information of the object is associated with the traffic instruction, the image merging system merges 1908 at least one portion of the received LiDAR information with at least one pixel associated with the received object information to generate merged information representing the object. In an embodiment, the image merging system filters the merged information based on the categorization information. In an embodiment, merging the at least one portion of the received LiDAR information with the at least one pixel associated with the received object information comprises merging a plurality of LiDAR points within a bounding box of the camera information,

[0198] The image merging system determines 1910 a location of the object relative to the vehicle based on the merged information representing the object. In an embodiment, the image merging system determines an orientation of the object based on the merged information.

[0199] In an embodiment, the image merging system operates 1912 the vehicle based on the location of the object. In an embodiment, the image merging system causes a separate controller of the vehicle to operate the vehicle based on the location of the object. In some examples, the image merging system notifies a vehicle controller that a traffic light is approaching with a red traffic signal (e.g., stop signal), and in response, the vehicle is controlled to stop the vehicle before the object (e.g., within a distance from the object).

[0200] In an embodiment, the image merging system assigns an instance identifier to the object based on the merged information (e.g., object #1, object #2, etc.). In an embodiment, the image merging system determines that the object is already associated with an instance identifier. In an embodiment, if the image merging system determines that the object is not associated with an instance identifier, tire image merging system assigns an instance identifier to the object.

[0201] In an embodiment, the image merging system determines if the object represents two distinct objects based on the merged information. In accordance with determining if the object represents two distinct objects, the image merging system assigns a unique instance identifier to each of the two distinct objects.

[0202] In an embodiment, the image merging system determines an accuracy of the object based on the merged information.

[0203] In an embodiment, the image merging system annotates a map based on the merged information. In an embodiment, annotating a map includes transmitting the location of the object, the classification information of the object, an instance identifier of the object, and / or a date of observing the object to a database hosting the map. In an embodiment, the image merging system updates an existing instance of the object on a map based on the merged information. In an embodiment, the image merging system determines when the location or orientation of the object changes and removes the location and / or orientation information associated with the object from the map. .In some examples, when a vehicle returns to a particular area, of the environment, the image merging system determines that a previously detected traffic light is no longer present. In some cases, the image merging system transmits instructions to delete the traffic light from the map or database.

[0204] In an embodiment, the image merging system determines at least one geometric feature of the object based on the merged information. In an embodiment, determining the at least one geometric feature of the object includes determining at least one edge of the object and at least one surface of the object. In an embodiment, determining the at least one geometric feature of the object includes determining a size of the object. In some examples, an edge detection is performed on the merged information to determine an edge of the object and the orientation of the edge is used to infer an orientation of the object.

[0205] In an embodiment, the at least one camera acquires the camera information and the at least one LiDAR device acquires the LiDAR information concurrently. In an embodiment, the at least one camera acquires the camera information after the at least one LiDAR device acquires the LiDAR information. In an embodiment, a timing difference between when the at least one camera acquires the camera information and the at least one LiDAR device acquires the LiDAR information is based on a velocity of the vehicle. 10206] In an embodiment, the image merging system receives updated LiDAR information from the at least one LiDAR device. For example, in an embodiment, updated LiDAR information is generated to confirm or reassess the position of the object that was previously detected. In an embodiment, acquiring updated LiDAR information is performed when the previous LiDAR information was acquired using less LiDAR points than a number of LiDAR points associated with the updated LiDAR information. In some examples, LiDAR information is acquired a second time to increase resolution in areas of the environment where a traffic instruction is likely (e.g,, above the vehicle, on the sidewalks, etc.). In an embodiment, acquiring updated LiDAR information is performed when the previous LiDAR information was acquired using less than a full field of view of the LiDAR system (e.g., less than a 360 degree view around the vehicle) than an updated field of view associated with the updated LiDAR information,

[0207] In an embodiment, updated LiDAR information is acquired based on a previously detected object. In some examples, the previously detected object was detected based on a previous LiDAR. information and previously merged information. In some examples, the updated LiDAR information is acquired when an accuracy of the object from previous merged information is below a threshold accuracy,

[0208] In an embodiment, the image merging system receives vehicle location information from at least one sensor of the vehicle wherein the vehicle location information includes a latitude and longitude of the vehicle. In this embodiment, a latitude and longitude of the object is determined based on the latitude and longitude of the vehicle. For example, using the current position of the vehicle, the image merging system determines the location in global coordinates of the objects and record the location in the map, database, or server.

[0209] In the foregoing description, embodiments of the invention have been described with reference to numerous specific details that may vary from implementation to implementation. The description and drawings are, accordingly, to be regarded in an illustrative rather than a restrictive sense. The sole and exclusive indicator of the scope of the invention, and what is intended by the applicants to be the scope of the invention, is the literal and equivalent scope of the set of claims that issue from this application, in the specific form in which such claims issue, including any subsequent correction. Any definitions expressly set forth herein for terms contained in such claims shall govern the meaning of such terms as used in the claims. In addition, when we use the term “further comprising,” in the foregoing description or following claims, what follows this phrase can be an additional step or entity, or a sub-step / sub-entity of a previously-recited step or entity. 05 03 25

Claims

1. A vehicle, comprising:at least one non-transitory computer-readable medium storing computer-executable instructions;at least one processor communicatively coupled to the at least one non-transitory computer-readable medium, the at least one processor configured to execute the computer executable instructions, the execution carrying out operations including:receiving LiDAR information from at least one LiDAR device;receiving object information of an object proximate to the vehicle based on camera information from at least one camera, the object information comprising at least one pixel of the camera information representing the object and the object information comprising categorization information representing a classification of the object;merging at least one portion of the received LiDAR information with at least one pixel associated with the received object information to generate merged information representing the object;determining a location of the object relative to the vehicle based on the merged information representing the object;determining an accuracy of one or more parameters of the object based on the merged information representing the object, the one or more parameters of the object comprising one or more of orientation, heading, and position;determining a confidence level of the merged information actually representing the object based on the accuracy of the one or more parameters of the object, the confidence level representing how confident the vehicle is that the object proximate to the vehicle is in fact the object actually represented by the merged information; anda control circuit communicatively coupled to the at least one processor, the control circuit configured to operate the vehicle based on (i) the location of the object relative to the vehicle and (ii) the confidence level of the merged information actually representing the object.

2. The vehicle of claim 1, wherein the at least one processor is further configured to carry out operations including: filtering the merged information based on the categorization information.

3. The vehicle of any of the preceding claims, wherein the at least one processor is further configured to carry out operations including: determining an orientation of the object based on the merged information.

4. The vehicle of any of the preceding claims, wherein the at least one processor is further configured to carry out operations including determining if the categorization information of the object is associated with a traffic instruction.

5. The vehicle of claim 4, wherein the traffic instruction is a traffic light or a traffic sign.

6. The vehicle of claim 4 or claim 5, wherein the at least one processor is further configuredto carry out operations including determining a traffic signal of the traffic instruction based on the merged information.

7. The vehicle of claim 6, wherein the control circuit is further configured to operate the vehicle based on the traffic signal of the traffic instruction.

8. The vehicle of any of the preceding claims, wherein the at least one processor is further configured to carry out operations including: assigning an instance identifier to the object based on the merged information.

9. The vehicle of claim 8, wherein the at least one processor is further configured to carry out operations including:determining if the object represents two distinct objects based on the merged information; andin accordance with determining if the object represents two distinct objects, assigning a unique instance identifier to each of the two distinct objects.

10. The vehicle of any of the preceding claims, wherein the at least one processor is furtherconfigured to carry out operations including: determining whether the merged information is reliable based on at least one of the location of the object, the categorization information of the object, or an instance identifier of the object.

11. The vehicle of any of the preceding claims, wherein the at least one processor is further configured to carry out operations including: annotating a map based on the merged information.

12. The vehicle of claim 11, wherein annotating a map comprises transmitting the location of the object, the categorization information of the object, an instance identifier of the object, and a date of observing the object to a database hosting the map.

13. The vehicle of claim 12, wherein annotating a map further comprises: transmitting at least one of a number of times that the object has been observed, a frequency that the object has been observed, or a date and time of observing the object to the database for determining the reliability of the merged information.

14. The vehicle of any of the preceding claims, wherein the at least one processor is further configured to carry out operations including: updating an existing instance of the object on a map based on the merged information.

15. The vehicle of any of the preceding claims, wherein the at least one processor is further configured to carry out operations including: determining at least one geometric feature of the object based on the merged information.

16. The vehicle of claim 15, wherein determining the at least one geometric feature of the object comprises determining at least one edge of the object and at least one surface of the object.

17. The vehicle of claim 15, wherein determining the at least one geometric feature of the object comprises determining a size of the object.

18. The vehicle of any of the preceding claims, wherein the at least one camera acquires thecamera information and the at least one LiDAR device acquires the LiDAR information concurrently.

19. The vehicle of any of claims 1-17, wherein the at least one camera acquires the camera information after the at least one LiDAR device acquires the LiDAR information.

20. The vehicle of claim 19, wherein a timing difference between when the at least one camera acquires the camera information and the at least one LiDAR device acquires the LiDAR information is based on a velocity of the vehicle.

21. The vehicle of any of the preceding claims, wherein merging the at least one portion of the received LiDAR information with the at least one pixel associated with the received object information comprises merging a plurality of LiDAR points within a bounding box of the camera information.

22. The vehicle of any of the preceding claims, wherein determining the confidence level of the merged information actually representing the object comprises determining the confidence level further based on additional information that comprises at least one of:a number of times the object has been observed,a number of times the object has been observed within a predetermined number of days, a date and time of observing the object, oran observation frequency of the object.

23. The vehicle of claim 1, wherein the at least one processor is further configured to: in response to determine that at least one of the accuracy or the confidence level is below a corresponding threshold, determining that a second look is warranted and controlling or instructing to control at least one of:a zoom feature of the at least one camera or a second camera system to zoom-inon the object,panning an area of the object based on at least one of the camera information or the merged information,a zoom feature of the at least one LiDAR device or a second LiDAR system to zoom-in on the object, orpanning an area of the object based on at least one of the LiDAR information or the merged information.

24. A method comprising:receiving LiDAR information from at least one LiDAR device of a vehicle;receiving object information of an object proximate to the vehicle based on camera information from at least one camera, the object information comprising at least one pixel of the camera information representing the object and the object information comprising categorization information representing a classification of the object;merging at least one portion of the received LiDAR information with at least one pixel associated with the received object information to generate merged information representing the object;determining a location of the object relative to the vehicle based on the merged information representing the object;determining an accuracy of one or more parameters of the object based on the merged information representing the object, the one or more parameters of the object comprising one or more of orientation, heading, and position;determining a confidence level of the merged information actually representing the object based on the accuracy of the one or more parameters of the object, the confidence level representing how confident the vehicle is that the object proximate to the vehicle is in fact the object actually represented by the merged information; andoperating the vehicle based on (i) the location of the object and (ii) the confidence level of the merged information actually representing the object.

25. A non-transitory computer-readable storage medium comprising at least one program for execution by at least one processor of a first device, the at least one program including instructions which, when executed by at least one processor, cause the first device to perform the method of claim 24.

Images