Pseudorange estimation from passive sensors

Abstract

The present invention relates to pseudorange estimation from passive sensors. A rough initial estimate of line of sight ("pseudorange") is generated substantially immediately after a passive sensor on a vehicle detects an object. The data is combined with a priori detection likelihood and a priori performance model of the sensor. These comparisons generate a posterior probability distribution of pseudorange estimates. Pseudorange estimates are derived from the probability distribution and output for use in detection and avoidance decisions and action planning. Pseudorange estimates can be updated to improve their accuracy, such as by using a recursive filter (e.g., a Kalman filter). In addition to the spatial and temporal position of the vehicle, other information (such as current atmospheric data) or known (or possible) activity of the vehicle at the current time in the area can be used to improve accuracy.

Description

Technical Field

[0001] The present disclosure relates generally to vehicles, and in particular, to a system and method for estimating pseudoranges from passive sensors to detected unknown objects. Background Art

[0002] Advances in real-time software, sensor technology, sensor fusion, etc. have made autonomous transportation operation not only a reality, but actually better (on average) than manual operation. Research and development is ongoing to enable autonomous operation of a wide variety of vehicles (including ground vehicles, aircraft, trains, ships, submarines, spacecraft, etc.).

[0003] The ability to detect (and in many cases avoid) a vehicle or other objects in the environment is a common requirement for automating the operation of many types of vehicles. This detect and avoid (DAA) capability is particularly pressing for autonomous aerial vehicles (commonly referred to as "drones"), which must be able to detect and avoid other aerial vehicles and potential conflicting objects during both the flight and ground movement phases of operation.

[0004] Once the autonomous vehicle performs the "detection" portion of DAA, a piece of information that is critical to the "avoidance" portion is the line of sight (LoS) range to the detected object, or the distance to the detected object. Active ranging technologies such as RADAR and LIDAR are well known. Active systems transmit electromagnetic (EM) radiation, and use the reflected transmitted radiation to determine an estimate of the target range through a series of signal processing techniques.

[0005] Active ranging systems are large and complex, and the EM transmitters and processing circuits required for processing signals and calculating distances all require a lot of power. In applications where weight and power are very valuable, it is desirable to use passive sensors for range detection. In addition, active transmission systems effectively broadcast the position of aircraft, which may be undesirable for national defense, law enforcement, or resident protection applications. As used herein, passive sensors are sensors that detect physical properties without sending any signals. Passive sensors can detect electromagnetic radiation in any spectrum portion (including visible light, infrared, radio frequency (RF) etc.). Passive sensors also include those sensors that detect audible sounds or ultrasonic sounds, or generally include any discernible physical properties representing the presence of unknown objects (such as, another vehicle).

[0006] Several techniques for distance detection using passive sensors are known in the art. Most existing solutions require: processing target information through multiple frames; additional hardware (i.e., installation of multiple sensors); images of the detected object with discernible features (e.g., measurable number of pixels along a particular dimension, return signal strength, discernible features, etc.); or multiple images and / or times from objects of different geometries (e.g., maneuvering of an autonomous vehicle to form an artificial baseline between images, pitch sensors, etc.). Each of these approaches exhibits shortcomings in their utility for DAA in autonomous vehicles (especially, light aircraft).

[0007] Model-based approaches to passive range detection use some discernible and measurable characteristics of the detection target (such as the number of pixels) as an estimate of the wingspan or fuselage size, color, style, identification marks, or other discernible features (such as the number of engines). These features are compared with known characteristics of possible aircraft and other objects to determine an estimate of the distance. Such methods have the potential to provide instantaneous distance measurements and generally do not require additional sensors. The disadvantage of methods that rely on discernible features of the target is that for all encounter geometries except a few potential encounter geometries, the discernible features of the detection object are generally indistinguishable until long after the initial detection. Therefore, even if there is sufficient time to make decisions and avoidance actions, an accurate range estimate may not be obtained. Model-based approaches are also sensitive to noise / error when estimating features (e.g., number of pixels, area, length of size, etc.), which generally deteriorates with increasing distance. In addition, model-based approaches rely on the characteristics of the assumed target (model), which may not be correct for encountering a given object, resulting in errors in the distance estimate. Distance measurements at perceptible distances require higher resolution sensors at the expense of reduced field of view. Multiple sensors are often required to cover the minimum effective field of view, adding to the size, weight, drag, and power requirements of the DAA system. Higher resolution sensors also require the use of more sophisticated image stabilization techniques.

[0008] A variation of the model-based approach measures the growth rate of some discernible feature of the target image or signal (e.g., pixel area, dimension, contrast, return intensity, etc.) and compares it to the rate of a hypothesized model, which can be used to estimate the range of the object based on the distance or time of the method's closest point. The growth rate method suffers from the same problems as other techniques, namely, relying on the tracking of discernible features of the target. In addition, the apparent growth of the discernible features typically follows an exponential profile, which causes the range estimate to occur too late in closed scenarios to be useful for DAA. The method is also sensitive to noise / error in estimating features (e.g., number of pixels, area, length of dimension, etc.), which typically deteriorates with increasing distance. The method relies on an assumed growth model, which in turn assumes certain characteristics of the target aircraft, and therefore the distance of detection depends on the accuracy of the initial classification of the detected object.

[0009] A well-known method of estimating distance is by using multiple spatially distributed sensors, known in the art as stereo sensing. Working in much the same way as the human eye, the distance to the target is determined from the parallax (differential image) generated between two or more images of a single target. The range finding performance is limited by the size and geometry of the baseline separating the sensors, the accuracy of the time synchronization of the images, and the ability to stabilize the images to avoid vibrations, etc. This method requires the installation of multiple sensors at spatially distributed locations of the autonomous vehicle, which increases drag and weight, increases power consumption, and reduces the amount of available payload. The range finding accuracy is proportional to the size of the baseline between consecutive images.

[0010] Another method of distance detection similar to stereo sensing is motion baseline sensing. In this method, sensor motion (through manipulation of autonomous vehicles or sensors themselves) can be used to provide multiple observations of the detection object in different relative postures. Similar image processing (as used in stereo sensing) is then applied to determine the range estimate of the target. The motion baseline method relies on accurate tracking of sensor position and posture (or their estimation through filtering) between images to generate distance estimates. A model of the target aircraft motion (between observations) is also required. The accuracy is proportional to the size of the baseline between consecutive images and the accuracy of the target aircraft motion model. A significant disadvantage of this method is that it takes additional time to collect and process multiple images of sufficiently different postures. Instantaneous estimation of distance is impossible, and the time spent on manipulation reduces the time available to handle the situation and take evasive action.

[0011] Another method of passive ranging uses motion parallax, occlusion, and relative size. This method is based on comparing the size, position, and / or motion of the detected object with other objects of known size and / or a distance from the sensor. In addition, the degree of overlap or occlusion of the target with other objects (within a known range) can also be used to estimate the distance to the target.

[0012] Although conceptually feasible, parallax and occlusion based methods may not be practical for DAA of autonomous light aircraft, as they require multiple "known and ranged" objects in the same scene as (or close to) the target of interest. These methods may not be as accurate as other ranging methods, providing "order of magnitude" estimates at best. These methods are most suitable for DAA systems used during ground operations (such as taxiing).

[0013] Known methods of distance estimation using passive sensors exhibit a number of drawbacks. The need for multiple sensors for stereo perception adds weight, drag, and power consumption to the autonomous vehicle. The need for sensor movement (either individually or by moving the vehicle) further increases complexity and is too slow to be used for DAA. Model-based growth rate and similar methods depend on early target classification and comparison to model errors stored in the initial classification, which is not conducive to accurate distance estimation. In addition, the distance estimates that these methods can produce are too late to be used in DAA. Parallax and occlusion-based methods depend on operation in an environment with known reference objects, which are generally unlikely to occur during autonomous vehicle operation.

[0014] Therefore, a range estimation technique using a single (or a few) passive sensors that obtains the closest range estimate, which does not rely on target feature detection or target classification, would represent a significant improvement over the state of the art.

[0015] The background section of this document is provided to place aspects of the present disclosure in a technical and operating context, thereby helping those skilled in the art understand their scope and utility. Unless expressly identified as such, statements herein are not admitted to be prior art merely by their inclusion in the background section. Summary of the invention

[0016] A brief summary of the present disclosure is given below to provide a basic understanding for those skilled in the art. This summary is not an extensive overview of the present disclosure and is not intended to identify key / important elements of aspects of the present disclosure or to describe the scope of the present disclosure. The sole purpose of this summary is to introduce some concepts disclosed herein in a simplified form as a preface to a more detailed description presented later.

[0017] According to one or more aspects described and claimed herein, a pseudorange is estimated after a passive sensor detects an object (a "pseudorange" is an estimate of the actual distance that is rough or known to be inaccurate). The detection data from the sensor is combined with a priori performance model of the sensor. These comparisons generate a posterior probability time of the pseudorange estimate, or such a distribution set. The pseudorange estimate is derived from the probability distribution and output for use in making DAA decisions and action planning. For example, the most likely pseudorange estimate, or the worst-case pseudorange estimate, can be selected. Basically, a pseudorange estimate is generated immediately after the passive sensor detects the object (although the accuracy may not be high). The pseudorange estimate can be updated to improve its accuracy, such as by using a recursive filter (e.g., a Kalman filter). Pseudorange estimates can also be improved using one or more known passive ranging methods. Combining the sensor detection data with the prior performance model of the sensor may include combining the spatial and temporal position of the sensor detection and the vehicle with a priori probability distribution describing the uncertainty in the sensor detection based on the pseudorange to the detected object using Bayesian techniques. Establishing multiple pseudorange estimates may include establishing a Bayesian posterior distribution of pseudoranges to unknown objects at the time of initial detection. In either or both of comparing to a priori performance models and deriving pseudorange estimates from a posterior probability distribution, other information in addition to the spatial and temporal position of the vehicle may be used, such as current atmospheric data or known (or possible) vehicle activity in the area at the current time.

[0018] One aspect relates to a method of estimating a pseudorange from a vehicle having a passive sensor to a detected unknown object. An indication of an object detection is received from a passive sensor. One or more probability distributions for pseudorange estimates are established by combining sensor detection data with a priori detection likelihood and a priori performance model of the sensor. An estimate of a pseudorange to the detected unknown object is derived from the one or more probability distributions for pseudorange estimates. An estimated pseudorange to the detected unknown object is output.

[0019] Another aspect relates to a pseudorange estimation device suitable for estimating a pseudorange from a vehicle having a passive sensor to a detected unknown object. The device includes a priori performance model stored for the sensor, a memory, and a processing circuit. The processing circuit is operably connected to the memory and is operable to receive the priori performance model. The processing circuit is suitable for: receiving data representing the detection of an object from a passive sensor; establishing one or more probability distributions of pseudorange estimates by combining the sensor detection data with the priori detection likelihood and the priori performance model of the sensor; deriving an estimate of the pseudorange of the detected unknown object from the one or more probability distributions of the pseudorange estimate; and outputting an estimated pseudorange to the detected unknown object. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The present disclosure will now be described more fully below with reference to the accompanying drawings, in which aspects of the present disclosure are shown. However, the present disclosure should not be understood to be limited to the aspects set forth herein. Rather, these aspects are provided to make the present disclosure more comprehensive and complete, and to fully convey the scope of the present disclosure to those skilled in the art. Throughout the text, similar reference numerals refer to similar elements.

[0021] Figure 1 is a block diagram of a DAA system including a pseudorange estimation system.

[0022] Figure 2 is a flow chart of a method of estimating a pseudorange from a vehicle having a passive sensor to a detected unknown object.

[0023] Figure 3 is realized Figure 1 Block diagram of an apparatus of a pseudorange estimation system. DETAILED DESCRIPTION

[0024] For the purpose of simplicity and illustration, the present disclosure is described by referring primarily to its exemplary aspects. In the following description, many specific details are set forth to provide a thorough understanding of the present disclosure. However, it will be readily apparent to one of ordinary skill in the art that aspects of the present disclosure may be practiced without limitation to these specific details. In this description, well-known methods and structures are not described in detail in order to avoid unnecessarily obscuring the present disclosure.

[0025] As used herein, the terms "pseudorange" and "pseudorange estimate" refer to an initial rough estimate of the LoS range or distance from the sensor to the object that is presented substantially immediately upon detection of the object by a passive sensor. The modifier "pseudo" indicates that the pseudorange estimate is not accurate enough for, for example, detailed flight planning. However, the pseudorange estimate is generally accurate enough to initiate the DAA process. "Substantially immediately" means that the pseudorange estimate is generated after only a small amount of storage access and computational operation delays when an indication of the detection of the object is received. Given the processing speed of modern processors and DSPs, this short delay is substantially instantaneous compared to known passive detection systems (which may require a large amount of processing time to track the target through several image frames or track the development of the detected characteristics, or may incur delays in moving vehicles to produce different sensing views). The initial detection of an object by a sensor (or circuit processing the output of the sensor) is generally only a binary indication of the presence of the object. At this stage, nothing is known about the specific details of the object (such as size, characteristics, color, etc.). Therefore, the initially detected object is referred to herein as a "detected unknown object" that has been detected, but its classification or identity is unknown.

[0026] The pseudorange estimation process of aspects of the present disclosure simulates the way humans intuitively pilot and, based on experience, make an initial estimate of the distance once aircraft, birds, and other objects in their airspace are visually detected. The process involves applying the experience stored in the pilot's memory using heuristic rules that are also generated and refined through experience. The main factors used in establishing the initial pseudorange estimate are the position of the aircraft carrying the sensor, the position of the detected object (e.g., another aircraft, etc.), and the time of day.

[0027] As a non-limiting example, a pilot flying at an altitude of 3,000 feet near a small non-towered airport can visually detect the reflection of some unknown objects "flicker" or sunlight from the nearby airspace. However, due to its distance, relative geometry, atmospheric visibility, etc., the pilot cannot clearly see the object or recognize its characteristics. Initially, the pilot attributes the flicker to an aircraft based on his or her experience (birds do not produce such reflections). Due to the spatial position at a relatively low altitude near a small airport, the pilot will further assume that the aircraft is probably a small piston aircraft. Based on previous experience (encoded as an implicit model of his or her own detection performance), the pilot knows that this aircraft will be clearly visible within about 2 miles, and it is difficult or impossible to see above about 6 miles, so a possible distance of 2-6 miles can be assumed. On the contrary, the same pilot flying at 12,000 feet near the main airport in Class B airspace may interpret similar visual flicker as the most likely commercial jet aircraft with more different visibility characteristics, thus assuming that the pseudo-range to the object is larger. In making these initial pseudorange estimates, the pilot instinctively and often subconsciously uses other available information. For example, current atmospheric visibility will affect the distance at which the pilot knows he or she can detect aircraft of various sizes, and he or she will adjust the pseudorange estimate accordingly. Similarly, human pilots naturally take into account their knowledge of known activities (such as training operations at a military base or air shows at nearby airports) when estimating the pseudorange to the detected unknown object. Of course, the pilot does not explicitly consider all of the above factors when making the distance estimate to the object. In addition, depending on the proximity of the object at the first visual detection, the pilot may not follow the considered or discussed distance analysis process. In contrast, automatic responses based on rough visual data can be all automatic responses that the pilot uses to initiate avoidance actions.

[0028] According to aspects of the present disclosure, after a passive sensor is instructed to detect an object, processing circuitry in an autonomous vehicle applies a similar reasoning process. For any particular type / size / sensitivity of sensor (e.g., visible light, IR, UV, etc.), multiple prior performance models are developed or obtained from actual experiments. For each of the possible distances to the object, the prior performance model relates to the LoS range from the sensor to the object under specific conditions (e.g., location, time, atmospheric conditions, etc.). These are encoded as prior probabilities and likelihood distributions. These prior performance models are stored in a database such as (e.g., a database implemented within a Bayesian belief network) and are available by training a neural network.

[0029] When a passive sensor indicates a detection of an object (without further details, such as discernible features of the object), a Bayesian updating process is used to combine the detection with a hypothesis likelihood or a sampling model with a prior (past) performance model of the sensor to determine a posterior probability distribution of pseudorange estimates for a set of assumed target types and encounter conditions. Typically, additional information (such as current atmospheric conditions or known activity patterns in the area, time of day, etc.) may also be considered as parameters or marginal probabilities (encounter conditions) in the likelihood model. Typically, by assuming different target types and encounter conditions, the process results in a set of pseudorange probability distributions for each different (possible) target type and condition.

[0030] In one aspect, Bayesian probability theory is used to generate a posterior probability distribution (or a collection of such distributions) for pseudorange estimates based on a comparison of current conditions with a prior performance model. Bayesian probability is a mathematical technique for making decisions based on imperfect or incomplete information. When information is not available, Bayesian analysis makes mathematical assumptions about what it may be. As this information is collected and disseminated, the Bayesian method corrects or permutes the assumptions and changes its decisions accordingly.

[0031] More formally, Bayesian inference techniques combine a prior probability distribution over the uncertainty quantity (e.g., pseudoranges to detected objects) with a likelihood function (e.g., a prior performance model for a particular target type and conditions such as spatial and temporal position, atmospheric conditions, etc.) to generate a posterior probability distribution over the uncertainty quantity (i.e., pseudoranges), given the information resulting from the comparison.

[0032] Various mathematical techniques can be used to derive an estimate of the LoS range from the output a posteriori pseudorange probability distribution. For example, the most likely pseudorange, or the worst case (typically, closest) one can be selected. If necessary, a model and / or a recursive filter (such as a Kalman filter) that describes the closing speed of the detected object in the past can be used to provide updates to the pseudorange estimate. In some aspects, prior art ranging techniques can also be used to increase the accuracy of the pseudorange estimate. As described herein, these may require the use of alternative or additional sensors, multiple frames of object data, movement of sensors, etc., which may require time to develop and process. At the same time, the initial pseudorange estimate can be used by the DAA process to potentially initiate an avoidance maneuver.

[0033] In some respects, as more accurate range information becomes available, the process may update its a priori performance models and marginal probabilities in a manner similar to how human pilots use their experience to enhance their own performance.

[0034] Figure 1 is a block diagram showing how pseudorange estimation according to aspects of the present disclosure may be incorporated into a general DAA system. The DAA system 10 receives input from a passive sensor 14. Typically, the DAA system 10 includes a detection unit 20, a tracking unit 22, an evaluation unit 24, an alarm unit 26, and an avoidance unit 28. One or more DAA units 20-28 may be implemented as discrete hardware blocks, such as separate processors, field programmable gate arrays, full custom hardware, etc. Alternatively, one or more DAA units 20-28 may be implemented as software modules executed on one or more processors, digital signal processors (DSPs), etc. In any specific embodiment, the functions of two or more DAA units 20-28 may be combined. The DAA units 20-28 perform the functions suggested by their names. Simply put, the detection unit 20 monitors the output signals of the passive sensor 14 and detects the presence of an object. The tracking unit 22 tracks the position of the detected object over time. When receiving the pseudorange estimation estimate, the evaluation unit 24 evaluates the threat of collision or interference with the detected object. If such a threat is high enough, the alarm unit 26 generates an alarm. After receiving the alert, the avoidance unit 28 plans one or more strategies to avoid the object, such as changing the route of the vehicle. Generally, the DAA function is known to those skilled in the art and will not be described in detail here.

[0035] The DAA system 10 also includes a pseudorange estimation system 12. The pseudorange estimation system 12 is operable to generate a pseudorange estimate substantially immediately after the passive sensor 14 detects an object as determined by the detection unit 20. The pseudorange estimation system 12 includes a stored a priori performance model 30, a Bayesian update processing unit 32, and an inference processing unit 34. In some aspects, the pseudorange estimation system 12 also includes a stored closure model 38 and a pseudorange update processing unit 36. The stored models 30, 38 may be in the form of a database (e.g., a database implemented within a Bayesian belief network), or may be loaded into a connection of a neural network through training. One or more processing units 32, 34, 36 may be implemented as discrete hardware blocks, such as separate processors, field programmable gate arrays, full custom hardware, etc. Alternatively, one or more processing units 32, 34, 36 may be implemented as software modules executed on one or more processors, digital signal processors (DSPs), etc. In any specific implementation, the functionality of two or more processing units 32, 34, 36 may be merged. The functions of the processing units 32, 34, 36 are discussed below, including the information input to each unit 32, 34, 36 and the information generated as output of each unit.

[0036] The Bayesian update processing unit 32 receives detection indications of objects from the passive sensor 14 via the detection unit 20. In some aspects (e.g., by Figure 1 ), the Bayesian update processing unit 32 may also receive information from the detection unit 20 regarding the orientation of the object (i.e., its direction relative to the passive sensor 14). Typically, the data indicating the detection of an object is simply a binary indicator of whether the object has been detected (1) or has not been detected (0). At this point, no detailed information about the detected object is available, or information required to generate a pseudorange estimate. For example, the size, shape, color, or features of the object need not be included in the detection indication. Nor is any necessary information available regarding the motion of the object over time, or the rate of increase of any detectable features, or any information regarding the detected object to nearby known objects. If any such data is available, the system can utilize such data, potentially achieving greater accuracy, however, in its broadest aspect, the object detection indication from the passive sensor 14 is simply a binary indication.

[0037] The Bayesian update processing unit 32 receives the prior and marginal probability distributions describing the uncertainty in the sensor detections from the prior performance model 30 based on the pseudorange to the detected object. The Bayesian update processing unit 32 combines the sensor detections with the prior performance model 30 of the sensor and outputs one or more posterior probability distributions of the pseudorange estimates (i.e., in some aspects, it outputs a set of probability distributions if different target types and conditions are encoded in the prior performance model 30). In one aspect, the one or more posterior probability distributions are generated using known Bayesian inference techniques and equations.

[0038] In certain aspects, the Bayesian update processing unit 32 may also receive current atmospheric condition data from various sources 16, such as weather, precipitation, cloud cover, visibility, temperature, and expected values ​​(indicative of fog), etc. In these aspects, at least some of the a priori performance models 30 also include atmospheric condition information. That is, the performance of the passive sensors 14 in varying atmospheric conditions that were previously simulated or recorded, and a comparison of this past performance data with the current atmospheric conditions may produce a more accurate posterior probability distribution for the pseudorange estimates.

[0039] In certain aspects, the Bayesian update processing unit 32 may also receive information related to known or possible vehicular activity in the area of ​​the spatial and temporal location of the vehicle from various sources 16. In certain aspects of these aspects, at least some of the a priori performance models 30 also include information related to known or possible vehicular activity. That is, the performance of the passive sensor 14 in the presence of different levels and different types of vehicular activity is previously simulated or recorded, and comparison of this past performance data with current known or possible vehicular activity and the current spatial and temporal location of the passive sensor 14 can produce a more accurate posterior probability distribution for the pseudorange estimate.

[0040] use Figure 1 and Figure 3 The dashed lines in delineate boxes representing the various sources 16 and their inputs to the Bayesian update processing unit 32, which boxes indicate that these inputs are optional and not necessary in the broadest concept of the present disclosure.

[0041] For example, an autonomous aircraft may routinely fly near a certain airport during the delivery of packages or other activities. If it is known that a skydiving school uses the airport as a venue on weekends, then the sensor's ability to detect aircraft of a particular size or type used during skydiving operations may be simulated or measured, and at least some of the a priori performance models 30 may include this information. In this case, a set of posterior probability distributions for pseudorange estimates may be obtained, some of which are based on the types of aircraft commonly used in skydiving operations. Based on other available information such as the spatial and temporal location of the passive sensor 14 (e.g., near skydiving activities and on weekends), and other possible data such as atmospheric conditions (e.g., skydiving activities are likely to be on sunny days rather than on rainy days), it is likely that the a priori probability distribution generated by the a priori performance model 30 for skydiving-type aircraft can be determined and can be selected to produce a more accurate pseudorange estimate. It should be noted that the factoring of additional information (such as vehicle location and weather) does not impose significant delays or require accurate characteristics of the detection object. Instead, because the additional information is encoded into some a priori performance model 30, a set of posterior probability distributions for pseudorange estimates can be generated just as quickly (or nearly as quickly) as a single probability distribution, with the only additional step being selecting a distribution from the set based on current position, conditions, or other information.

[0042] The posterior probability distribution of the pseudorange estimate (or set of such distributions) is transmitted to the inference processing unit 34, which derives therefrom a pseudorange estimate for the detected unknown object. The pseudorange is output to the evaluation unit 24 in certain aspects (as shown by the language in brackets between the inference process 34 and the evaluation 24) as well as an indication of the type of object. In one aspect, the inference processing unit 34 derives the pseudorange estimate by selecting the pseudorange estimate with the highest probability of being correct from one or more posterior probability distributions. In another aspect, the inference processing unit 34 selects the worst-case pseudorange from the one or more posterior probability distributions. Typically, the worst-case pseudorange will be the shortest such estimate. In one aspect (not in Figure 1 ), the inference processing unit 34 also receives at least some information from various sources 16, such as vehicle location, current atmospheric conditions, etc. For example, the inference processing unit 34 may output the most likely pseudorange estimate for a clear day in one or more posterior probability distributions; however, during times of low visibility, a worst-case pseudorange estimate may be output.

[0043] In some respects, such as through Figure 1As shown by the dashed lines in the figure, the pseudorange estimates provided to the DAA system 10 can be continuously updated. In these aspects, the inference processing unit 34 also provides one or more posterior probability distributions of the pseudorange estimates to the pseudorange update processing unit 36, and the one or more posterior probability distributions can continuously change due to changes in inputs to the Bayesian update processing unit 32 over time. The pseudorange update processing unit 36 ​​applies a temporary state estimator to the posterior probability distribution. An example of such a temporary state estimator is a Kalman filter, which is a known tool for predicting the behavior of a dynamic system in the face of uncertain information based on past behavior. The pseudorange update processing unit 36 ​​provides updated (i.e., evolving over time) pseudorange estimates to the evaluation unit 24 of the DAA system 10, and in some aspects provides (as shown by the language in brackets between the pseudorange update processing 36 and the evaluation 24) additional updated object type estimates.

[0044] Figure 2 A method 100 of estimating a pseudorange from a vehicle having a passive sensor 14 to a detected unknown object is depicted. Data representing object detection is received from the passive sensor 14 (block 102). In one aspect, the data is a binary indication. One or more probability distributions for pseudorange estimates are established by combining the sensor detection data with an a priori detection performance model 30 (block 104). An estimate of a pseudorange to the detected unknown object is derived from the one or more probability distributions for the pseudorange estimates (block 106). The estimated pseudorange to the detected unknown object is output (block 108) to an evaluation unit 24 of the DAA system 10. In certain aspects ( Figure 2 The method 100 may provide pseudoranges to detected unknown objects for use in, for example, the DAA system 10, by combining own aircraft position data (e.g., from a navigation system) with pseudorange estimates to provide a position estimate of the target object in a navigation coordinate frame (e.g., latitude, longitude, altitude).

[0045] In one aspect, deriving a pseudorange estimate for the detected unknown object from one or more probability distributions of pseudorange estimates includes selecting a pseudorange estimate with a high probability of being correct. This can provide accurate estimates for efficient DAA operation.

[0046] In another aspect, deriving a pseudorange estimate for the detected unknown object from the one or more probability distributions of pseudorange estimates includes selecting a worst case pseudorange estimate. This can ensure that the object is avoided in a timely manner out of caution.

[0047] In one aspect, combining the sensor detection data with the a priori performance model 30 of the sensor 14 includes combining the sensor detection data with a priori probability distribution describing uncertainty in sensor detection based on pseudoranges to the detected objects using Bayesian techniques. Further, in one aspect, establishing a probability distribution for pseudorange estimates includes establishing a Bayesian posterior probability distribution of pseudoranges to the position objects for each initial contact. This generates pseudorange estimates using known mathematical techniques.

[0048] In one aspect, the method 100 further comprises receiving current atmospheric condition data, and wherein one or both of the combining step and the deriving step further comprises additionally considering the atmospheric condition data. In one aspect, the method 100 further comprises accessing a database of known or possible vehicle activities in the area of ​​the vehicle's spatial and temporal position, and wherein one or both of the combining step and the deriving step further comprises additionally considering the vehicle's activities. The incorporation of additional environmental conditions or activities may improve the accuracy of the pseudorange estimate.

[0049] In one aspect, the method 100 further includes updating the estimated pseudorange to the detected unknown object by comparing the previous pseudorange estimate to one or more current probability distributions of pseudorange estimates using the temporary state estimator. In one aspect, the temporary state estimator further receives a previous closed model 28 of the distance of the detected object. This can provide increasingly accurate pseudorange estimates over time.

[0050] In one aspect, a probability distribution of pseudorange estimates is established for hypothesized detected object types and environmental conditions. This increases the range of possible pseudorange estimates from which to choose and can improve accuracy.

[0051] Figure 3 Describes an implementation according to one aspect of the present disclosure Figure 1 The passive sensor 14 and various sources 16 as well as the target detection unit 20 and evaluation unit 24 of the DAA system 10 are depicted to put the pseudorange estimation system 12 in context. These devices and functional units are described above with respect to Figure 114. The pseudorange estimation system 12 includes a processing circuit 40, a memory 42, and at least an a priori performance model 30. As shown by brackets, the pseudorange estimation system 12 may additionally include a closed model 38. The processing circuit 40 is suitable for implementing the method 100. Specifically, the processing circuit 40 is suitable for receiving data representing detection of an object from a passive sensor; establishing one or more probability distributions of pseudorange estimates by combining the sensor detection data with the a priori detection likelihood of the passive sensor 14 and the a priori performance model 30; deriving an estimate of the pseudorange of the detected unknown object from the one or more probability distributions of the pseudorange estimate; and outputting the estimated pseudorange of the detected unknown object. This can provide the DAA system 10 with essentially instantaneous pseudorange estimates, from which the DAA system can assess hazards and plan avoidance procedures.

[0052] In one aspect, the processing circuit 40 is adapted to derive a pseudorange estimate for the detected unknown object from one or more probability distributions of pseudorange estimates by selecting a pseudorange estimate with a high probability of being correct. This may provide a more accurate distance for efficient DAA operation.

[0053] In another aspect, the processing circuit 40 is adapted to derive a pseudorange estimate for the detected unknown object from one or more probability distributions of pseudorange estimates by selecting a worst case pseudorange estimate. This can ensure that the object is avoided in a timely manner as a precaution.

[0054] In one aspect, the processing circuit 40 is adapted to combine the sensor detection data with an a priori probability distribution describing uncertainty in the sensor detection according to the pseudorange to the detected object by using Bayesian techniques to combine the sensor detection data with the a priori performance model 30 of the sensor. Further, in one aspect, the processing circuit 40 is adapted to establish a probability distribution of pseudorange estimates by establishing a Bayesian posterior probability distribution of pseudoranges to unknown objects at initial contact. This generates pseudorange estimates using known mathematical techniques.

[0055] In one aspect, the processing circuit 40 is further adapted to receive current atmospheric condition data, and wherein the processing circuit 40 is adapted to perform one or both of the combining step and the deriving step by additionally taking into account the atmospheric condition data. In one aspect, the processing circuit 40 is further adapted to access a database of known or possible vehicle activities in the space-time position area of ​​the vehicle, and wherein the processing circuit is adapted to perform one or both of the combining step and the deriving step by additionally taking into account the vehicle activities. The incorporation of additional environmental conditions or activities may improve the accuracy of the pseudorange estimate.

[0056] In one aspect, the processing circuit 40 is further adapted to update the estimated pseudorange of the detected unknown object by comparing the previous pseudorange estimate with one or more current probability distributions of pseudorange estimates using a temporary state estimator. In one aspect, the processing circuit 40 further receives a previous closed model 38 of the distance of the detected object. This can provide increasingly accurate pseudorange estimates over time.

[0057] In one aspect, the processing circuit 40 is adapted to establish a probability distribution of pseudorange estimates for assumed detected object types and environmental conditions. This increases the range of possible pseudorange estimates from which to select and may improve accuracy.

[0058] The processing circuit 40 may include any one or more sequential state machines (one or more sequential state machines operable to execute machine instructions stored in the memory 42 as a machine-readable computer program), such as one or more hardware-implemented state machines (e.g., discrete logic, FPGA, ASIC, etc.); programmable logic together with appropriate firmware; one or more stored program, general-purpose processors, such as a microprocessor or a digital signal processor (DSP); or any combination of the above. Although the memory 42 is depicted as being separate from the processing circuit 40, those skilled in the art understand that the processing circuit 40 includes internal memory, such as a cache memory or a register file. Those skilled in the art further understand that virtualization techniques allow some functions nominally performed by the processing circuit 40 to be actually performed by other hardware, perhaps remotely located (e.g., in the so-called "cloud") . In such aspects, the pseudorange estimation system 12 (or other vehicle electronic components) may additionally include one or more wireless communication interfaces (not shown), such as a cellular or WLAN interface.

[0059] The memory 42 may include any non-transitory machine-readable medium known or developable in the art, including but not limited to magnetic media (e.g., floppy disks, hard drives, etc.), optical media (e.g., CD-ROMs, DVD-ROMs, etc.), solid-state media (e.g., SRAM, DRAM, DDRAM, ROM, PROM, EPROM, flash memory, solid-state disks, etc.), etc.

[0060] The a priori performance model 30 and closed model 38 (if present) may be stored in one or more databases or other memory structures, or may be loaded into a connection of one or more neural networks through training operations as is known in the art.

[0061] exist Figure 3 In the aspects described in Figure 1 Depicted and referenced in Figure 1The depicted Bayesian update processing unit 32 and inference processing unit 34 are implemented as software modules stored in memory 42 and executed by processing circuitry 40. In one aspect, memory 42 additionally stores, and processing circuitry 40 additionally executes, a pseudorange update processing unit 36, which may implement a temporary state estimator, such as a Kalman filter, to provide updated pseudorange estimates.

[0062] Aspects of the present disclosure have many advantages over the passive ranging systems of the prior art. Aspects are operable to provide an estimate of the pseudorange to the detected object substantially immediately after receiving a binary indication of the detection of the object. No object data from multiple frames over time is required. No different object maps from different sensor positions are required. No additional known objects are required. There is no need to resolve the discernible features of the detected object, or to characterize / assume its size, shape, etc. in an early stage. No model that does not conform to the assumed spectral characteristics is required. Aspects of the present disclosure provide an estimate of the pseudorange to the detected unknown object based only on the detection indication from the passive sensor in one aspect, the spatial and temporal position of the vehicle, and the prior performance model of the sensor. In other aspects, additional information such as atmospheric conditions or known activities can improve the accuracy of the pseudorange estimate. Since only separate passive sensors and electronic circuits are required, the pseudorange estimation system according to aspects of the present disclosure minimizes the weight, drag, complexity, and power consumption of autonomous vehicles (such as, light aircraft). Compared with the prior art passive object detection system, this saves energy for useful payloads and expands the range of autonomous small aircraft. Although known to be of limited accuracy, the generated pseudorange estimate is accurate enough to provide the DAA system 10 with an early useful rough estimate of the object's LoS distance to assess the situation and potentially initiate early avoidance procedures.

[0063] Furthermore, the present disclosure includes embodiments according to the following items:

[0064] Item 1. A method for estimating a pseudorange from a vehicle having a passive sensor to a detected unknown object, comprising the following steps:

[0065] receiving data representing detection of an object from a passive sensor;

[0066] establishing one or more probability distributions for pseudorange estimates by combining the sensor detection data with a priori detection likelihood and a priori performance model of the passive sensor;

[0067] deriving an estimate of a pseudorange for a detected unknown object from the one or more probability distributions of pseudorange estimates; and

[0068] Output is the estimated pseudorange to the detected unknown object.

[0069] Item 2. A method according to item 1, wherein deriving an estimate of the pseudorange to the detected unknown object from one or more probability distributions of pseudorange estimates includes selecting a pseudorange estimate with a high probability of being correct.

[0070] Item 3. A method according to item 1 or 2, wherein deriving an estimate of the pseudorange to the detected unknown object from one or more probability distributions of pseudorange estimates includes selecting a worst-case pseudorange estimate.

[0071] Item 4. A method according to any one of Items 1-3, wherein combining the sensor detection data with the prior performance model of the sensor includes using Bayesian techniques to combine the sensor detection data with a prior probability distribution describing uncertainty in the sensor detection based on the pseudorange to the detection object.

[0072] Item 5. A method according to any of Items 1-4, wherein establishing one or more probability distributions for pseudorange estimates includes establishing a Bayesian posterior probability distribution of pseudoranges to unknown objects at each initial contact.

[0073] Item 6. The method according to any one of items 1 to 5, further comprising:

[0074] Current atmospheric condition data is received, and wherein one or both of the combining step and the deriving step further comprises additionally taking into account the atmospheric condition data.

[0075] Item 7. The method according to any one of Items 1 to 6, further comprising:

[0076] A database of known or possible vehicle activity in the area of ​​the vehicle's spatial and temporal location is accessed, and wherein one or both of the combining and deriving steps further include additionally considering the vehicle's activity.

[0077] Item 8. The method according to any one of Items 1 to 7, further comprising:

[0078] An estimated pseudorange to the detected unknown object is updated by comparing a previous pseudorange estimate to one or more current probability distributions of pseudorange estimates using a temporary state estimator.

[0079] Item 9. A method according to any one of Items 1-8, wherein the temporary state estimator also receives a previous closed model of the range of the detected object.

[0080] Item 10. A method according to any one of Items 1-9, wherein a probability distribution of pseudorange estimates is established for assumed detection object types and environmental conditions.

[0081] Item 11. A pseudorange estimation apparatus adapted to estimate a pseudorange from a vehicle having a passive sensor to a detected unknown object, comprising:

[0082] A priori performance models for passive sensor storage;

[0083] Memory; and

[0084] A processing circuit operatively connected to the memory and operable to receive the a priori performance model, the processing circuit being adapted to perform the following steps:

[0085] receiving data representing detection of an object from a passive sensor;

[0086] establishing one or more probability distributions for pseudorange estimates by combining the sensor detection data with a priori detection likelihood and a priori performance model of the passive sensor;

[0087] deriving an estimate of a pseudorange for a detected unknown object from the one or more probability distributions of pseudorange estimates; and

[0088] Output is the estimated pseudorange to the detected unknown object.

[0089] Item 12. An apparatus according to item 11, wherein the processing circuit is adapted to derive an estimate of the pseudorange of the detected unknown object from the one or more probability distributions of pseudorange estimates by selecting a pseudorange estimate with a high probability of being correct.

[0090] Item 13. An apparatus according to Item 11 or 12, wherein the processing circuit is adapted to derive an estimate of the pseudorange of the detected unknown object from one or more probability distributions of the pseudorange estimates by selecting a worst-case pseudorange estimate.

[0091] Item 14. An apparatus according to any one of Items 11-13, wherein the processing circuit is suitable for combining the sensor detection data with a prior performance model of the sensor by combining the sensor detection data with a prior probability distribution describing the uncertainty in the sensor detection based on the pseudorange to the detection object using Bayesian techniques.

[0092] Item 15. The apparatus of Item 14, wherein the processing circuit is adapted to establish one or more probability distributions of pseudorange estimates by establishing a Bayesian a posteriori probability distribution of pseudoranges to the unknown object at each initial contact.

[0093] Item 16. An apparatus according to any one of Items 11-15, wherein the processing circuit is further adapted to receive current atmospheric condition data, and wherein the processing circuit is adapted to perform one or both of the combining step and the deriving step by additionally taking into account the atmospheric condition data.

[0094] Item 17. An apparatus according to any one of Items 11-16, wherein the processing circuit is further adapted to access a database of known or possible vehicle activities in the area of ​​the vehicle's spatial and temporal location, and wherein the processing circuit is adapted to perform one or both of the combining step and the deriving step by additionally considering the vehicle activities.

[0095] Item 18. An apparatus according to any of Items 11-17, wherein the processing circuit is further adapted to update the estimated pseudorange to the detected unknown object by comparing a previous pseudorange estimate with one or more current probability distributions of pseudorange estimates using a temporary state estimator.

[0096] Item 19. An apparatus according to any one of Items 11-18, wherein the processing circuit also receives a previous closed model of the range of the detection object.

[0097] Item 20. An apparatus according to any of Items 11-19, wherein the processing circuit is suitable for establishing a probability distribution of pseudorange estimates for assumed detected object types and environmental conditions.

[0098] Although aspects of the present disclosure are discussed herein primarily with reference to autonomous aircraft operations, those skilled in the art will readily recognize that all of these aspects are well suited for many applications of autonomous operation of vehicles and are not limited to aircraft.

[0099] Of course, without departing from the essential features of the present disclosure, the present disclosure may be implemented in other ways than those specifically set forth herein. The proposed aspects should be considered in all respects as illustrative and not restrictive, and all changes falling within the meaning and equivalent scope of the appended claims are intended to be included in the appended claims.

Claims

1. A method for estimating a pseudorange from a vehicle having a passive sensor to a detected unknown object, comprising the following steps: receiving passive sensor detection data representing detection of an object from the passive sensor, wherein the passive sensor detection data is a binary indication of the presence of the object; establishing one or more probability distributions for pseudorange estimates by combining passive sensor detection data with a priori detection likelihood and a priori performance model of the passive sensor using Bayesian techniques; deriving an estimate of the pseudorange of the detected unknown object from the one or more probability distributions of the pseudorange estimates by selecting a pseudorange estimate with a high probability of being correct or selecting a worst-case pseudorange estimate; and Output is the estimated pseudorange to the detected unknown object.

2. The method according to claim 1, wherein: Combining the passive sensor detection data with an a priori performance model of the sensor includes combining the passive sensor detection data with an a priori probability distribution describing uncertainty in sensor detection based on pseudoranges to detected objects using Bayesian techniques.

3. The method according to claim 2, wherein: Establishing the one or more probability distributions for the pseudorange estimates includes establishing a Bayesian posterior probability distribution of the pseudorange to the unknown object at each initial contact.

4. The method according to claim 1, further comprising: An estimated pseudorange to the detected unknown object is updated by comparing a previous pseudorange estimate to one or more current probability distributions of pseudorange estimates using a temporary state estimator.

5. The method according to claim 4, wherein: The interim state estimator also receives a previous closed model of the range of the detected object.

6. The method according to claim 1, wherein: A probability distribution of pseudorange estimates is established for the assumed detected object type and environmental conditions.

7. A pseudorange estimation device adapted to estimate the pseudorange from a vehicle having a passive sensor to a detected unknown object, comprising: a priori performance model stored for the passive sensor; Memory; as well as a processing circuit operatively connected to the memory and operable to receive the a priori performance model, the processing circuit being adapted to perform the following steps: receiving passive sensor detection data representing detection of an object from the passive sensor, wherein the passive sensor detection data is a binary indication of the presence of the object; establishing one or more probability distributions for pseudorange estimates by combining passive sensor detection data with a priori detection likelihood and a priori performance model of the passive sensor using Bayesian techniques; deriving an estimate of the pseudorange of the detected unknown object from the one or more probability distributions of the pseudorange estimates by selecting a pseudorange estimate with a high probability of being correct or selecting a worst-case pseudorange estimate; and The estimated pseudorange to the detected unknown object is output.

8. The device according to claim 7, wherein: The processing circuit is adapted to combine the passive sensor detection data with an a priori performance model of the sensor by combining the passive sensor detection data with an a priori probability distribution describing uncertainty in sensor detection according to pseudoranges to detection objects using Bayesian techniques.

9. The device according to claim 8, wherein: The processing circuit is adapted to establish one or more probability distributions of the pseudorange estimates by establishing a Bayesian a posteriori probability distribution of pseudoranges to unknown objects at each initial contact.

10. The device according to claim 7, wherein: The processing circuit is further adapted to update the estimated pseudorange to the detected unknown object by comparing a previous pseudorange estimate to one or more current probability distributions of pseudorange estimates using the temporary state estimator.

11. The device according to claim 10, wherein: The processing circuit also receives a previous occlusion model of the range of the detected object.

12. The device according to claim 8, wherein: The processing circuit is adapted to establish one or more probability distributions of the pseudorange estimates for assumed detected object types and environmental conditions.

Images