JP7152395B2 - Gap measurement for vehicle platoons - Google Patents

Description

Cross-reference to related applications

This application is based on U.S. patent application Ser. No. 1, pp. 100-120, which is hereby incorporated by reference in its entirety.

The present invention relates generally to systems and methods for allowing vehicles to safely tandem in close proximity to each other using automatic or partially automatic control.

Recent years have seen significant progress in the field of autonomous and semi-autonomous vehicles. One segment of vehicle automation relates to vehicle platooning systems that allow vehicles to be lined up in close proximity in a safe, efficient and convenient manner. Staying close behind another vehicle can have great fuel-saving advantages, but is generally unsafe when done manually by the driver. One type of vehicle platoon system is sometimes referred to as a vehicle platoon system in which a second and potentially additional vehicle is automatically or semi-automatically controlled to safely align closely with the preceding vehicle. .

In vehicle platoon and platoon systems, understanding the distance between vehicles becomes a very important control parameter, and several different independent mechanisms can be used to determine the distance between vehicles. This may include radar systems, transmission of absolute or relative position data (eg, GPS or other GNSS data) between vehicles, LIDAR systems, cameras, and the like. A challenge that arises when using radar in platoon-type applications is the need to reliably identify a partner vehicle from a series of potentially ambiguous radar returns and track it under constantly changing conditions. This application describes techniques for identifying and tracking specific vehicles based on vehicle radar data that are well suited for platoons, platoons, and other automated or semi-automated driving applications.

Various methods, controllers, and algorithms are described for identifying and/or tracking the rear of a particular vehicle (e.g., a platoon partner) in a set of ranging scenes. there is The techniques described can be used with a variety of different ranging technologies including radar, LIDAR, sonar units or any other time-of-flight ranging sensors, camera-based ranging units, and the like.

In one aspect, radar (or other ranging) scenes are received and the relative position of each detected object they represent, and in some circumstances the relative velocity of such detected objects, are determined. , with the estimated position (and relative velocity) for the first vehicle, at least partially identifying the first vehicle point candidate. The first candidate vehicle points are classified based on their respective distances from the estimated location of the first vehicle to the detected object. The classification is repeated for multiple samples such that the first classified vehicle point candidates include candidates from multiple consecutive samples. A rear portion of the first vehicle is then identified based at least in part on the classification of the first candidate vehicle point. Subsequently, the identified rear portion of the first vehicle, or an effective vehicle length determined based at least in part on the identified rear portion of the first vehicle, can be used to control the second vehicle. .

In some embodiments, a bounding box is conceptually applied around the estimated location of the first vehicle, and measurement system object points not located within the bounding box are not considered first candidate vehicle points. not done. In some embodiments, the bounding box defines an area that exceeds the maximum expected size of the first vehicle.

In some embodiments, the vehicle's relative velocity is estimated with an associated velocity uncertainty. In such an embodiment, an object point of the collection of detected object points moving with a relative velocity not within the velocity uncertainty of the estimated velocity is the first candidate vehicle point. is not considered.

In some embodiments, classifying the first candidate vehicle point includes populating a histogram with the first candidate vehicle point. The histogram includes a plurality of bins, each bin representing a longitudinal distance range for the estimated position of the first vehicle. In such an embodiment, after the histogram contains at least a predetermined number of candidate first vehicle points, identification of the rear of the first vehicle may be performed. In some embodiments, a clustering algorithm (eg, a modified mean shift algorithm, etc.) is applied to the first candidate vehicle points to identify one or more clusters of the first candidate vehicle points. In such an embodiment, the rear of the first vehicle is selected by selecting the cluster located closest to the second vehicle that includes at least a predetermined threshold percentage or a predetermined number of first vehicle radar point candidates. can be represented.

In some embodiments, Kalman filtering is used to estimate the position of the first vehicle.

In another aspect, a method of tracking a particular leading vehicle using a distance measuring unit onboard a trailing vehicle is described. In this embodiment, current radar (or other ranging) samples are obtained from the radar (or other ranging) unit. The current rangefinder sample contains a collection of zero or more object points. In parallel, a current estimate of the state of the preceding vehicle corresponding to the current sample is obtained. The current state estimate may include a position parameter (such as the current relative position of the preceding vehicle), a speed parameter (such as the current relative speed of the preceding vehicle), and/or other position and/or orientation related parameters, provided that , including but not limited to one or more state parameters.

The current estimate of the state of the preceding vehicle has an associated state uncertainty and does not take into account any information from the current range measurement samples. Within the state uncertainty, a determination is made as to whether any of the object points match the estimated state of the preceding vehicle. If there is a match, the matching object point that best matches the estimated state of the preceding vehicle is selected as the measured state of the preceding vehicle. That measured state of the preceding vehicle is then used to determine the next sequential estimate of the state of the preceding vehicle corresponding to the next sequential sample. By repeating the above steps multiple times, the preceding vehicle is tracked. The measured state of the leading vehicle is used in controlling one or both of the vehicles, e.g. can be used in at least partially automatic control of

In some embodiments, each sample indicates, for each object point, the position of the detected object (relative to the range finding unit) corresponding to such object point. Each current estimate of the state of the preceding vehicle includes a current estimate of the (relative) position of the preceding vehicle and has an associated position uncertainty. To be considered a valid measurement, the selected matching object point must match the estimated position of the preceding vehicle within the position uncertainty. In some implementations, the estimated current position of the leading vehicle estimates the current position of the rear of the leading vehicle.

In some implementations, each sample indicates, for each object point, the relative velocity of the detected object (relative to the range measuring unit) corresponding to such object point. Each current estimate of the state of the preceding vehicle includes a current estimate of the relative speed of the preceding vehicle and has an associated speed uncertainty. To be considered a valid measurement, the matching object point selected must match the estimated relative velocity of the preceding vehicle within the velocity uncertainty.

In some embodiments, if none of the radar object points in a particular range measurement sample match the estimated state of the preceding vehicle within the state uncertainty, then the estimation of the state of the next preceding vehicle in sequence is performed. State uncertainty increases.

In some embodiments, global positioning satellite system (GNSS) position updates are received periodically based at least in part on the detected GNSS positions of leading and trailing vehicles. Each time a vehicle GNSS position update is received, the estimated state and state uncertainty of the preceding vehicle are updated based on such position update.

In some embodiments, vehicle speed updates are received periodically based at least in part on the detected wheel speeds of the preceding and following vehicles. Each time a vehicle speed update is received, the estimated state and state uncertainty of the preceding vehicle are updated based on such preceding vehicle speed update.

In another aspect, various methods, controllers, and algorithms are described for fusing sensor data obtained from different vehicles for use in at least partially automated control of a particular vehicle. The described technique is well suited for use with a variety of different vehicle control applications, including platoons, platoons, and other connected driving applications.

In one aspect, the information about the second vehicle is obtained at the first vehicle using a first sensor on the first vehicle while the first vehicle and the second vehicle are driving. detected. Information regarding the second vehicle is also received by the first vehicle from the second vehicle. Utilizing the received information about the second vehicle helps determine whether the sensed information about the second vehicle is a valid measurement of the second vehicle. Subsequently, the first vehicle is automatically controlled, at least in part, based at least in part on aspects of the sensed information regarding the second vehicle.

In some embodiments, the first sensor measures the distance to the second vehicle. In some implementations, the first sensor also detects the speed of the second vehicle relative to the first vehicle. In different embodiments, the first sensor is a radar unit, a LIDAR unit, a sonar unit, a time-of-flight distance sensor, a sensor configured to receive signals transmitted from a beacon on the second vehicle, a camera, and It can be any of the stereo camera units.

In some embodiments, the second vehicle information received includes a global positioning satellite system (GNSS) position fix of the current position of the second vehicle, speed information indicating a speed or relative speed of the second vehicle (e.g. wheel speed, etc.), and an indication of at least one of acceleration, heading, steering angle, yaw rate, tilt, lean or lateral movement of the second vehicle.

In some embodiments, the second vehicle information received includes the predicted state of the second vehicle. The predicted state may optionally include one or more of predicted position, predicted velocity, predicted acceleration, predicted heading, predicted yaw rate, predicted tilt, predicted tilt, and predicted lateral movement of the second vehicle.

The described approach is well suited for use in vehicle platoon and/or vehicle platoon systems, including tractor-trailer truck platoon applications.

The invention and its advantages may best be understood by referring to the following description in conjunction with the accompanying drawings.

1 is a block diagram of a representative platoon control architecture; FIG. Fig. 4 is a flow chart illustrating a method for determining the effective length of a platoon partner based on radar unit output; FIG. 4 is a schematic diagram illustrating bounding box behavior for expected partner vehicle positions; FIG. 4 is a schematic diagram illustrating exemplary radar object points that may be identified by a radar unit associated with a trailing truck directly behind a leading truck; FIG. 4B is a schematic diagram illustrating a situation in which the entire leading truck of FIG. 4A is not within the field of view of the radar unit; 4B is a schematic diagram illustrating a situation where the bounding box associated with the leading track of FIG. 4A is not completely within the field of view of the radar unit; FIG. Fig. 3 is a schematic diagram showing a situation where the leading truck is in a different lane than the trailing truck, but its entire bounding box is within the field of view of the radar unit; FIG. 5A is a graph showing the relative positions (longitudinal and lateral) of a first representative cluster of potential partner vehicle radar points that may be detected when tethered to a tractor-trailer rig. FIG. 5B is a histogram representing the longitudinal distance of the detected partner vehicle radar point candidates shown in FIG. 5A. FIG. 5C is a plot showing the mean shift center of the histogram points represented in FIG. 5B. FIG. 5D is a graph showing the relative positions (longitudinal and lateral) of a second (expanded) set of candidate partner vehicle radar points that may be detected when tethered to a tractor-trailer rig. FIG. 5E is a histogram representing the longitudinal distance of the detected partner vehicle radar point candidates shown in FIG. 5D. FIG. 5F is a plot showing the mean shift center of the histogram points represented in FIG. 5E. 1 is a schematic block diagram of a radar scene processor suitable for use by a vehicle controller to interpret received radar scenes; FIG. Figure 7 is a flow chart illustrating a method for determining if any particular radar scene reports a position behind a partner vehicle and updating the estimates of Figure 6; FIG. 4 is an illustration showing a Kalman filter state array and covariance matrix suitable for use with some embodiments;

In the drawings, like reference numbers may be used to indicate like structural elements. It should also be understood that the depictions in the figures are schematic and not to scale.

The Applicant has proposed various vehicle platoon systems that automatically or semi-automatically control a second and potentially further vehicle to safely align closely with the preceding vehicle. For example, US Pat. Nos. 5,300,000 and 5,000,000 describe various vehicle platoon systems that are at least partially automatically controlled such that a trailing vehicle is closely lined up with a designated leading vehicle. Each of these earlier applications is incorporated herein by reference.

One of the purposes of the platoon is typically to maintain a desired longitudinal distance between platoon vehicles, often referred to herein as the "desired gap." That is, it is desirable for a trailing vehicle (eg, trailing truck) to maintain a specified gap relative to a particular vehicle (eg, leading truck). Vehicles involved in platoons typically have sophisticated control systems suitable for initiating the platoon, maintaining the gap under a variety of different driving conditions, and appropriately breaking the platoon when necessary.

The architecture and design of control systems suitable for implementing vehicle platoons can vary widely. For example, FIG. 1 diagrammatically illustrates a vehicle control architecture suitable for use in a tractor-trailer truck platoon. In the illustrated embodiment, the platoon controller 110 receives input from a plurality of sensors 130 or other connectivity units on the tractor and/or one or more trailers, and a plurality of actuators and an actuator controller 150 control the power of the tractor. configured to control the operation of trains and other vehicle systems; An actuator interface (not shown) may be provided to facilitate communication between platoon controller 110 and actuator controller 150 . Platoon controller 110 also interacts with vehicle-to-vehicle communication controller 170, which coordinates communications with platoon partners, and NOC communications controller 180, which coordinates communications with a network operations center (NOC). The vehicle also preferably has selected configuration files containing known information about the vehicle.

Some of the functional components of platoon controller 110 include gap regulator 112 , mass estimator 114 , radar tracker 116 and brake health monitor 118 . In many applications, platoon controller 110 includes various other components as well.

Some of the sensors utilized by platoon controller 110 are GNSS (GPS) unit 131, wheel speed sensor 132, inertial measurement unit 134, radar unit 137, LIDAR unit 138, camera 139, accelerator pedal position sensor 141, steering wheel position. It may include sensor 142, brake pedal position sensor 143, and various accelerometers. Of course, not all of these sensors are available on all vehicles involved in the platoon, nor are all of these sensors required in certain embodiments. In other embodiments, various other sensors (either existing or later developed or commercially deployed) may be utilized additionally or alternatively by the platoon controller. In the main embodiment described herein, GPS location data is used. However, GPS is just one of the currently available global positioning satellite systems (GNSS). Therefore, it should be understood that data from any other GNSS system or other suitable location system may be used instead of, or in addition to, the GPS system.

Many (but not all) of the sensors described, including wheel speed sensor 132, radar unit 137, accelerator pedal position sensor 141, steering wheel position sensor 142, brake pedal position sensor 143, and accelerometer 144, pull the semitrailer. It is relatively standard equipment on newer trucks (tractors) used for However, others, such as the GNSS unit 131 and LIDAR unit 138 (if used), may not currently be standard equipment on such tractors or may not be present on certain vehicles, It can be installed as needed or desired to help support the platoon.

Some of the vehicle actuator controllers 150 that are at least partially directed by the platoon controller include a torque demand controller 152 (which may be integrated into the ECU or powertrain controller), a transmission controller 154, a brake controller 156 and a clutch controller 158. .

Communication between vehicles may be directed over any suitable channel and may be coordinated by vehicle-to-vehicle communication controller 170 . For example, the Dedicated Short Range Communication (DSRC) protocol (eg, the IEEE 802.11p protocol), which is a two-way short to medium range wireless communication technology developed for vehicle-to-vehicle communication, works well. Of course, other communication protocols and channels can be used in addition to or instead of the DSRC link. For example, vehicle-to-vehicle communication may additionally or alternatively be a CB radio channel, one or more GMRS bands, and one or more FRS bands, or any other existing radio channel using any suitable communication protocol. , or via a later developed communication channel.

The specific information exchanged between vehicles can vary widely based on the needs of the platoon controller. In various embodiments, the transmitted information may include current commands generated by the platoon controller such as requested/commanded engine torque, requested/commanded brake deceleration, and the like. When these aspects are controlled by a platoon controller, the information may also include steering commands, gear commands, and the like. Corresponding information indicates that these commands are controlled by a platoon controller or other automatic or semi-automatic controller (e.g. Adaptive Cruise Control System (ACC) or Collision Mitigation System (CMS)) on the partner vehicle, or other or more traditional Received from the partner vehicle, whether or not generated via a mechanism (eg, in response to driver input (eg, accelerator pedal position, brake position, steering wheel position, etc.)).

In many embodiments, most or all of the tractor sensor information provided to the platoon controller is also transmitted to the platoon partner, and corresponding information is received from the platoon partner, so that each vehicle's platoon controller can You can create an accurate model of what you are doing. The same is true for any other relevant information provided to the platoon controller, including any vehicle configuration information associated with the platoon controller. It should be understood that the specific information transmitted can vary widely based on the requirements of the platoon controller, the sensors and actuators available on each vehicle, and the specific knowledge each vehicle may have about itself.

Information transmitted between vehicles may also include information about intended future actions. For example, if the vehicle ahead knows it is approaching a hill, it is expected to increase its torque demand (or decrease its torque demand in downhill conditions) in the near future, and this information , to the following vehicle for use accordingly by the platoon controller. Of course, there is a wide variety of other information that can be used to predict future torque or brake demands, and that information can be communicated in a variety of different forms. In some embodiments, the nature of the expected event itself (eg, approaching a hill or curve or exit, etc.) can be indicated along with the expected timing of such event. In other embodiments, intended future operations can be reported in terms of expected control commands, such as expected torque and/or other control parameters, and when such changes are expected. Of course, there are various types of expected events that can be associated with platoon control.

Communications between the vehicle and the NOC may be transmitted over a variety of different networks such as cellular networks, various Wi-Fi networks, satellite communications networks, and/or various other networks as desired. . Communication with the NOC may be coordinated by NOC communication controller 180 . The information sent to and/or received from the NOC can vary widely based on the overall system design. In some situations, the NOC may provide specific control parameters such as target gap tolerances. These control parameters or constraints may be based on factors known to the NOC, such as speed limits, nature of the road/terrain (e.g. hilly vs. flat, curvy vs. straight, etc.), weather conditions, traffic conditions or road conditions. good. In other situations, the NOC can provide such information to the platoon controller. The NOC may also provide information about the partner vehicle, including its configuration information and any known pertinent information regarding its current operating conditions such as weight, trailer length, and the like.

Vehicles participating in the radar tracking platoon typically have one or more radar systems that are used to detect nearby objects. Since radar systems tend to be very good at determining distances between objects, the separation distance reported by radar units is very useful in controlling gaps between vehicles. Therefore, once a platoon partner has been identified, it is important to find that particular partner vehicle in light of the radar system's output. That is, it is important to determine which of the various different objects that may be identified by the radar unit (if any) matches the target partner.

As a preliminary, it should be understood that the platoon partner does not always correlate with the closest vehicle detected by the radar unit or the vehicle directly in front of the trailing truck. There are various different scenarios that can cause this case. For example, when the platoon is first installed, the partner may be too far away to be seen by the host vehicle's radar unit. Once a partner has sighted the radar unit, it becomes important to identify and distinguish that partner from other objects within the radar unit's field of view. The following description is specified from other objects that can be detected by the radar unit so that the radar unit can effectively track the partner vehicle (sometimes also referred to as "locking" to the partner). Techniques particularly well suited for identifying and distinguishing between partners are described.

Further, in the course of driving, there is traffic passing or overtaking the platoon in adjacent lanes passing next to the platoon so that the gap controller does not attempt to maintain a gap from the wrong vehicle. It is important that the unit can continue to distinguish from vehicles passing platoon partners. In another example, the leading truck may change lanes and may not be directly in front of the following vehicle at this time, so the distance between the platoon partners reported by the radar unit may be the most It is important to be associated with platoon partners, not with vehicles that are close or happen to be directly in front of the trailing truck. In some cases the radar unit cannot "see" the platoon partner. This may be due to an intruder breaking in between platoon partners, or the preceding vehicle being maneuvered out of the field of view of the following vehicle's radar unit, interference with the radar signal, or the like.

For platoon control purposes, it is also important to understand where the rear of the vehicle is relative to the vehicle's reported position. Specifically, the location of the partner vehicle is generally known from GPS-based location information transmitted to the host vehicle. However, GPS systems typically report positions on the tractor, such as the position of the antenna that receives the GPS signals. The detected GPS position is also translated to a reference position on the vehicle that is a known distance from the GPS antenna, and that reference position serves as the vehicle's reported GPS position. The particular reference position selected is variable based on control system preferences. For example, in some tractor-trailer truck platoon embodiments, the reference position may be the center of the rear axle of the tractor.

The difference between the reported GPS position and the physical rear of the vehicle can be important for platoon control. Therefore, it is often important to know the distance between the reported vehicle position and the actual rear of the vehicle. This is sometimes referred to herein as the "effective vehicle length". Effective vehicle length is particularly useful in tractor-trailer truck situations where the reported GPS position is typically somewhere in the cab (tractor) and the distance from the reported GPS position to the rear of the trailer is fairly long. is important. For example, trailer lengths on the order of 12-18 meters are common in the United States, although shorter or longer lengths (actually much longer for double or triple trailers) are possible. The distance from the reported GPS position to the rear of the vehicle must also take into account the longitudinal distance from the reported GPS position to the front of the trailer and/or any load-related extensions. In the trucking industry, any particular tractor may pull a variety of different trailers, and the effective vehicle length is often unknown because the attachment points between the tractor and trailer are adjustable on the tractor. should be understood.

Establishing a Radar Fix for Platoon Partners As is evident from the discussion above, the challenge that arises in using radar in Platoon type applications is to first locate and identify the partner vehicle against the output of the radar system, and then One is the need to reliably track under constantly changing conditions. In applications such as the trucking industry, it is also desirable to determine at least the effective length of the leading vehicle.

Commercially available radar units used in typical road vehicle automation systems typically provide data indicating the presence of any objects detected within a given field, along with the relative position and velocity of such objects. Output. Thus, while driving, such radar units may detect the presence of various objects within their field of operation. Detected objects include any vehicle directly in front of the host vehicle, vehicles in adjacent lanes that may pass, be overtaken by, or run alongside the platoon, and obstacles in the road. , signs, trees, and other roadside objects. Many different types of objects can be detected, but the radar units themselves typically do not know or convey the identity or nature of the detected objects. Rather, the radar unit simply reports the relative position and motion of any perceived object within its operating range. Therefore, in order to identify and track a partner vehicle in relation to the radar unit's output, the logic that interprets the radar unit's output must ensure that the partner vehicle is on the radar whether or not the partner vehicle is within its field of view. It is useful to accurately know and maintain the assumed location relative to the unit's line of sight. This is because the platoon system preferably has multiple independent mechanisms that can be used to help determine the vehicle's position, so no explicit mechanism is provided for identifying partners. It is possible even if

Once the platoon partners are identified, communication links are preferably established between the platoon vehicles. Communication can be established via one or more wireless links such as DSRC links, cellular links, and the like. Once communication is established between two vehicles, they each begin to send and receive data about themselves, ie their current position and operational status. The processes used to identify potential platoon partners, establish platoons and appropriate communication links can vary widely. For example, some representative techniques are described in previously filed US Pat. incorporated herein by reference.

Once the platoon partner is identified, the platoon controller 110 requests the radar system control logic to attempt to locate the partner vehicle. More specifically, the trailing vehicle's radar tracker 116 needs to locate and track the rear of the leading vehicle against the output of the radar unit so that the data can be used for gap control. A particularly suitable method for establishing a radar fix on a platoon partner will now be described with reference to FIG. One aspect of establishing a radar fix is determining the length of the partner so that the GPS position information can be correlated to the radar system output.

When the process begins, the radar tracker's control logic determines, receives or requests an estimate of the partner vehicle's current relative position, and the partner vehicle's relative position is available, as represented by step 203 of FIG. When it is, it will read or periodically receive updates on the relative position of the partner vehicle. In addition to relative position, the estimated information can optionally include various additional position-related information, such as relative velocity of the vehicle, relative heading of the vehicle, and the like.

In some embodiments, the radar tracker's control logic estimates the partner vehicle's current relative position, velocity, and heading (bearing) based on various sensor inputs from both the host and partner vehicles. configured as As noted above, platoon partners communicate with each other, and during a platoon they exchange extensive information about themselves, including constantly updated information regarding their current location and operational status. For example, some of the location-related information that can help interpret the radar unit data is the partner vehicle's GPS position, wheel speed, heading/heading (the direction the vehicle is pointing), yaw rate (indicating how fast the vehicle is turning) , pitch, roll, and acceleration/deceleration (longitudinal and angular in any of the above directions). Operation-related information may also include various other information of interest, such as current torque demand, brake input, gears, and the like. Information about the vehicle may include information such as the make and model of the vehicle, its length (if known), its equipment, estimated weight, and the like. Any of these and/or other available information can be used for location related estimation. By way of example, one particular position estimator is described below with respect to FIGS.

While a specific estimator is described, it is understood that the relevant information for the estimated partner vehicle position can be obtained from any suitable source and the estimation need not be made by the radar tracker control logic itself. want to be Furthermore, while it is preferred that position and motion information be transmitted in both directions between vehicles, it is not essential as long as the host vehicle can obtain the necessary information about the partner vehicle.

Location-related information is updated very frequently. Although the actual frequency of updates may vary widely based on the nature of the information being updated and the nature of the communication link or vehicle system providing the information, the number of items such as GPS position and wheel speed received over the DSRC link may vary. An update frequency works well for frequencies on the order of 10-500 Hz, for example 50 Hz, and in other embodiments slower and much faster update frequencies may be used as desired. Furthermore, although periodic updates of location-related information are desirable, they need not be received synchronously or at regular intervals.

It should be appreciated that the partner vehicle may or may not be within the field of view of the radar unit when the radar system attempts to locate the partner vehicle. However, since both the host and partner vehicle locations are generally known, at least based on the received GPS data, it is easy to estimate their separation with reasonable certainty. GPS position signals tend to be fairly good, but the reported position can be somewhat off, so rather than treating the reported position as foolproof information, some reported It should be appreciated that it is better to treat GPS position as a somewhat uncertain estimate. Further details regarding some specific algorithms suitable for estimating partner vehicle position are detailed further below. GPS position readings from commercial GPS sensors used for vehicle automation applications tend to be accurate to within about a few meters in real road conditions if they have line-of-sight to at least four GPS satellites. However, of course, some GPS sensors are generally more accurate, but due to variables such as interference and operating in areas without line-of-sight to the required number of GPS satellites, all GPS sensors are always accurate. is not guaranteed.

Once the partner vehicle's relative position estimate is known, a bounding box is applied around the partner's estimated relative position (step 206 in FIG. 2). The purpose of the bounding box is to define an area where partner vehicles are "expected" to be found. Logic then looks for radar-detected objects located within that bounding box to identify objects that may be correlated with the partner vehicle. The bounding box concept is useful for several reasons. First, it should be understood that GPS units generally report the position of the GPS unit's antenna, which in the case of a tractor-trailer truck is on the cab. Typically, this detected position is then transformed to a predetermined reference position on the tractor and that transformed position is used as the reported GPS position. The GPS position reported for the tractor-trailer is therefore well short of the rear of the trailer, which is (a) the most important point for gap control purposes, and (b) typically the trailing platoon. It is the most prominent feature identified by radar units from partners. Additionally, the distance between the reported GPS position and the rear of the trailer may not be discernible in many situations. One reason for the uncertainty is that a particular tractor (cab) can be used to pull a variety of different trailers (or other loads) of potentially different lengths. Therefore, the effective length of the tractor-trailer combination can change from trip to trip and from a control standpoint, relying on the driver to manually enter the effective length of the tractor-trailer combination for each trip. is generally undesirable. To a lesser extent, the reported GPS positions of both platoon partners are subject to some degree of uncertainty.

The actual size and geometry of the bounding box used may vary, but the full range of vehicle lengths and widths plus possible buffers is included to account for the uncertainty in the estimated GPS position. It is desirable that the area be large enough to Therefore, for trucking applications, it is desirable that the longitudinal length of the bounding box be longer than any tractor-trailer combination expected to be encountered. For example, commercial trucking applications in the United States, including conventional tractor-trailer combinations, generally do not significantly exceed 23 meters in total length. For such applications, it has been found that a bounding box that is 32 meters long and on the order of 3-4.5 meters wide, for example 3.8 meters wide, works well. In areas where longer trailers or double or triple trailers are available, the tractor-trailer combination may be longer and therefore a longer bounding box may be appropriate. If the actual length of the platoon partner is known, the size of the bounding box can be adjusted to more accurately reflect the expected deviation between the GPS position and the rear of the trailer. to the effective vehicle length. However, even when the effective length and width of the platoon partner are considered "know", the inaccuracy of GPS estimates and the nature of the cargo extending beyond the vehicle's reported length are not included. It may still be desirable to utilize a bounding box of size larger than the reported length and width to accommodate the possibility of obtaining.

The bounding box need not be essentially rectilinear, but the bounding box can encompass any desired geometric shape and/or dimensions other than longitudinal length and lateral width ( relative velocity). Accordingly, the bounding box can be defined in any desired manner.

A typical bounding box 255 applied around a leading track 251 in a two-track platoon is shown schematically in FIG. In the illustrated embodiment, each truck has a GPS unit 258 located on its tractor (cab) and a radar unit 260 located in the front of the cab. It can be seen that the bounding box exceeds the length and width of the preceding track 251 .

この場合、対象物は拒否される。このような手法では、バウンディングボックスは、第３の軸である速度を有する状態空間マップ内におけるチューブの外観を効果的に有する。そのような手法の論理は、検出された対象物の測定された横方向オフセットおよび測定された速度の両方が比較的低い確率の一致である場合、検出されたポイントは、これらのパラメータのうちの一方がずれているが、他方が想定値に非常に近接している場合よりも一致する可能性が低い（従って、パートナー車両の後部を識別するという目的のために無視するのにより適している）。いくつかの特定のバウンディングボックスを定義する手法のみを説明してきたが、他の実施形態では、必要に応じて多種多様な他のバウンディングボックスの定義を使用することができることは明白であるはずである。更に、バウンディングボックスの定義は、経時的に変化するように構成されてもよい。例えば、アルゴリズムにより、どのレーダー対象サンプルポイントがパートナー車両またはパートナー車両の後部に対応する可能性が高いかについての理解が深まり始めると、バウンディングボックスのうちの１つ以上の選択された寸法を縮小することができる。 In some embodiments, the bounding box may be defined more complicatedly. For example, in one particular embodiment, the square of the vehicle's lateral offset (Y _off ) and relative velocity (V) can be compared to a threshold (Th). Even if the radar point is within the vertical extent of the bounding box, if the sum of their squares exceeds a specified threshold (Th), the radar point is rejected. Such a test can be expressed mathematically as shown below.

In this case the object is rejected. In such an approach, the bounding box effectively has the appearance of a tube in the state-space map with the third axis, velocity. The logic of such an approach is that if both the measured lateral offset and the measured velocity of the detected object are relatively low-probability matches, the detected point is Less likely to match than if one is off but the other is very close to the expected value (thus more suitable to ignore for the purpose of identifying the rear of the partner vehicle) . Although only some particular bounding box definition techniques have been described, it should be apparent that other embodiments may use a wide variety of other bounding box definitions as desired. . Further, the bounding box definition may be configured to change over time. For example, as the algorithm begins to develop a better understanding of which radar target sample points are likely to correspond to the partner vehicle or the rear of the partner vehicle, it reduces one or more selected dimensions of the bounding box. be able to.

Once the bounding box is established, logic determines whether the entire bounding box is within the field of view 263 of the other vehicle's radar unit (step 209). If not, the logic waits for the entire bounding box to be within the field of view of the radar unit and usefully ignores the radar system output to identify the partner vehicle (of course, radar system output can also be used if desired). can also be used for other purposes such as collision avoidance). There are a variety of reasons why a partner vehicle may not be in view, or completely out of view, of a radar unit at any particular time. First, it should be appreciated that although the radar units used to support the platoon can be placed in a variety of different locations on the vehicle, they often have a relatively small field of view. For example, one common approach is to place a forward-looking radar unit with a relatively narrow fixed beam near the center of the front bumper to detect objects in front of the vehicle. Such an arrangement is shown in FIG. Also shown in the figure is the field of view 263 of the radar unit 260 located on the trailing truck 252 .

If a forward facing radar unit is used, it will not be able to see any vehicles behind or to the side of its host vehicle. Field of view when the partner vehicle is far ahead or around a corner of the host, such as when the platoon partner is first identified, even when the partner vehicle is ahead of the radar unit host may deviate from In some cases, the platoon partner may be partially within the radar unit's field of view. A common example of such cases is when the partner vehicle is in an adjacent lane and the rear of its trailer is not far enough forward to be recognized by a narrow, forward-facing radar unit. There are some cases. If the rear of the bounding box is not within the field of view of the radar unit, it is undesirable to use radar samples as there is a risk that the rearmost part of the partner vehicle detected by the radar unit is not actually the rear of the vehicle. Please understand.

Figures 4A-4D show a few (out of many) possible relative positions of two trucks in the process of establishing a platoon. In FIG. 4A, leading truck 251 is directly in front of trailing truck 252 and its bounding box 255 is completely within field of view 263 of trailing truck radar unit 260 . In contrast, in FIG. 4B, leading truck 251 is in a lane adjacent to trailing truck 252, and some, but not all, leading truck 251 itself (and thus not all bounding boxes 255) It is within the field of view 263 of the trailing truck radar unit 260 . In FIG. 4C , leading truck 251 is in the lane adjacent to trailing truck 252 , and all leading trucks 251 themselves, but not the entire bounding box 255 , are within field of view 263 of trailing truck radar unit 260 . In FIG. 4D, the leading truck 251 is in the lane adjacent to the trailing truck 252, but unlike FIGS. It is in. If the entire bounding box is not within the radar unit's field of view (eg, as shown in FIG. 4B or 4C, or if the preceding vehicle is otherwise not visible), the partner vehicle identification logic, at step 209, determines whether the bounding box Wait for the entire box to come into view of the radar unit.

If the entire bounding box is within the field of view of the radar unit (eg, scenarios such as those shown in FIG. 4A or FIG. 4D), the radar system controller logic obtains the next radar sample (step 212) and determines the partner vehicle's position and self. Get the current estimate of the velocity for (step 215). Commercially available short-range radar units utilized for road vehicle applications are typically configured to output their sensed scene at a relatively fast sample rate. Each scene typically identifies a set of zero or more objects detected and the velocity of such objects relative to the radar unit itself.

The nature of radar systems is that a transmitted radio wave can be reflected by almost anything in the radio's path, including both any intended target and potentially a wide variety of different items. Therefore, when attempting to establish a platoon, it is important to identify the reflected signal representing the desired partner so that the partner can be distinguished from noise reflected from other objects. For example, when driving along a road, the radar unit will monitor any vehicle directly in front, passing vehicles traveling in the same or opposite direction, highway or roadside signs, trees, or other objects along the road. Reflections from a number of different vehicles may be received, including side objects and the like.

Once the sensed scene is received, the radar system control logic determines whether any of the identified objects are partner vehicle radar point candidates, as represented by step 218 . Representative objects that may be detected by radar unit 260 are represented by X's in FIGS. 4A-4D. Objects detected in the scene must be within the bounding box in terms of both position and velocity in order to qualify as partner vehicle radar point candidates. Radar objects located outside the bounding box are preferably rejected because they are more likely not to correspond to a partner vehicle. For example, they may correspond to vehicles in adjacent lanes 272, 273, intruders (not shown) located between platoon partners, objects on the side of roadway 274, and the like. Objects that do not closely match the partner vehicle's assumed relative velocity are also rejected because they are unlikely to correspond to a platoon partner, even if they are longitudinally and laterally aligned with the expected position of the bounding box. preferably. For example, stationary objects such as features on the side of the road (e.g. road signs, trees or stationary vehicles), debris in the road, or features detected on the road itself (e.g. potholes, etc.) Appears to be approaching the radar unit at the speed at which it is traveling. Note that many commercially available radar units automatically filter out stationary objects and do not report them. When such radar units are used, stationary objects will not even be identified as part of the radar scene.

Some of the reported radar objects are is moving in the same direction as the host vehicle, but You may be moving at a different relative speed than your expected partner speed. Since it is relatively likely that such radar objects do not correspond to partner vehicles, These types of radar points are also preferably ignored.

All detected radar objects that appear to match the partner's expected position and expected velocity within the defined bounding box are considered partner vehicle radar point candidates, and they represent the partner's estimated position (e.g., partner (along the partner's longitudinal axis) from the GPS location of the partner). In some embodiments, histograms are utilized for this classification. The number of bins in the histogram is variable. For ease of calculation, we have found that dividing the 512 bins evenly over the length of the bounding box works well, but using more or fewer bins for any particular application can also In an implementation using a bounding box of approximately 32 meters with 512 bins, each bin corresponds to approximately 6 cm (2-3 inches). More bins can be used if higher resolution is desired.

A short-range radar unit utilized in road vehicle applications can actually detect multiple different "objects" that are part of the same vehicle, as represented by radar points 276-279 in FIGS. 4A-4D. It has been observed that it is common to identify . This is especially common in trucks, and indeed it is common for a tractor-trailer truck's radar signature to appear as more than one object. For example, if the rear of the trailer, the underride guard, and/or other features of the trailer or cargo located near the rear of the trailer appear in the radar output as one or more separate objects (e.g., points 276, 277). There is In addition, objects further above the trailer and/or objects near the cab may be separately identified (eg, points 278, 279). For example, if the radar is mounted relatively low on the host vehicle, transmission or other items along the truck undercarriage, or other features of the tractor-trailer such as the trailer's landing gears or the rear of the tractor. Reflections can be detected to identify those items as separately detected "objects." Therefore, any particular sample may identify multiple objects that meet the partner vehicle's radar point candidate criteria (in practice, this is relatively common). In such situations, multiple candidates associated with a particular radar sample are added to the histogram.

After populating the histogram with candidate partner vehicle radar points identified in the sample, a determination is made at step 224 as to whether sufficient samples have been taken to analyze the radar data and identify the partner vehicle. If not, the logic returns to step 212 where the next sample is obtained and the process repeats until enough samples have been obtained to facilitate analysis. If the bounding box is partially out of the radar unit's field of view at any point (as represented by the "no" branch out of decision block 225), the logic determines , return to step 209 to wait for the entire bounding box to come into view.

As mentioned above, commercially available short-range radar units used for road vehicle applications are typically configured to output the sensed scene at a relatively fast sample rate. For example, sample rates on the order of 20-25 Hertz are common, although higher or lower sample frequencies may be used. Therefore, the histogram fills up very quickly when the partner vehicle is within the field of view of the radar unit and provides a fairly good representation of the partner's radar signature.

FIG. 5A is a plot showing a set of 98 detected candidate partner vehicle radar points transposed into the reference frame based on the expected position of the truck ahead. The x-axis of the plot shows the vertical distance from the expected position ahead of the preceding track to the detected point. The y-axis also indicates the lateral offset of the detected point relative to the central axis of the preceding track. Although there is noticeable variation in the locations of the detected points, it can be seen that in the illustrated sample set the points tend to be clustered in some regions. FIG. 5B is a histogram showing the vertical distance to each partner vehicle radar point candidate detected in the plot of FIG. 5A. It can be seen that clustering tends to be more pronounced when only vertical distances are considered.

The large cluster 290 at the far end of the histogram typically corresponds to the rear of the vehicle and is often (but not always) the largest cluster. Further forward clusters 292 typically correspond to other features of the partner track. Studies have shown that radar returns from forward features tend to be weaker and more sporadic identified as discrete objects by radar units, which translates into smaller clusters in histograms. .

If sufficient samples were obtained to support the analysis, the logic follows the "yes" branch from decision block 224 to step 227 where a clustering algorithm is applied to the histogram data. The trigger point for initiating processing can vary widely based on the needs of any particular system. In general, it is desirable for the histogram to contain enough data points so that the partner vehicle can be accurately identified. In some particular embodiments, the histogram must include data from samples corresponding to a first threshold (e.g., data corresponding to at least 3 seconds or samples corresponding to 60 samples), and , must include partner vehicle radar point candidates that correspond to at least the second threshold (eg, at least 60 partner vehicle radar points). The threshold used is variable based on the needs of a particular implementation. For example, in some embodiments, at least 1-5 seconds worth of data or samples corresponding to a threshold in the range of 40-500 points may be used. In one particular example, at least 3 seconds worth of data or samples corresponding to 60 samples and 60 partner vehicle radar points is used as the threshold.

The data set shown in FIGS. 5A and 5B represents the data set available at the first attempt to identify the rear of the partner vehicle, ie, the first time the “yes” branch from step 224 is followed.

In general, clustering algorithms group together data points that are highly likely to represent the same point. Various conventional clustering algorithms can be used for this purpose. For example, the modified mean shift algorithm works well. FIG. 5C is a plot showing the mean shift center of the histogram points represented in FIG. 5B, with the height of the center indicating the number of points associated with that center. With this representation, two clusters 290 and 292 stand out even more dramatically.

The average shift data is then analyzed at step 230 to determine if one of the clusters meets predetermined criteria for the rear of the partner vehicle. If so, the cluster is identified as corresponding to the rear of the vehicle (step 233). Since each cluster corresponds to a specified distance between the partner's reported GPS position and the rear of the vehicle, the effective length of the vehicle is defined by the clusters. As noted above, the term "effective vehicle length" as used herein corresponds to the distance between the reported GPS position and the rear of the vehicle, which is important to know for control purposes. distance. It should be understood that this will usually differ from the actual length of the vehicle, as the reported reference position may not be located in front of the vehicle.

In some implementations, the cluster located closest to the rear of the bounding box that exceeds a threshold percentage of the total number of radar points in the histogram is identified as the rear of the platoon partner vehicle. In some implementations, an additional constraint is used that requires cluster positions not to move more than a certain threshold over the last sample. For example, for some applications a maximum movement threshold of around 1 mm has been found to work well. This approach very reliably identifies radar points corresponding to the rear of the truck, even if the radar unit controller has no pre-determined knowledge of the length of the vehicle and regardless of the presence of other traffic. It has been found. However, it should be understood that the threshold percentage or other characteristic of the histogram used to identify the rear of the vehicle can vary based on the application. In the embodiment shown in Figures 5A-5C, cluster 290 is designated as the back of the preceding track.

Even though other traffic paralleling the platoon may be detected by the radar, the described technique effectively applies many different types of filters to filter these radar points very reliably. particular attention should be paid to Radar points reporting features that are not where the platoon partner is predicted to be are filtered because they are not within the bounding box. Radar points not moving near the expected relative velocity are filtered regardless of where they are found. The vehicle rear criterion used for the clustered histogram data effectively filters out all other vehicles moving within the bounding box footprint at approximately the same speed as the platoon partner. Because the bins are small enough, it is highly likely that such an intruder could maintain a constant gap large enough to trick the algorithm into thinking that the intruder was part of the target. (e.g., if an intruder is traveling at approximately the same speed as the partner vehicle, but is within the bounding box, the position of the intruder relative to the different enough to fall). The vehicle's rear reference also eliminates the more random objects reported by the radar unit.

The effective vehicle length indicated by the selected mean shift cluster may be reported to the gap controller and any other controllers related to partner length. In most cases, the distance between the GPS reference position and the front of the host vehicle is known, so the effective vehicle length determined by the radar unit is the distance ahead and behind the truck, as represented by step 236. can easily be used in conjunction with known information about the track to positively indicate the

In some situations, none of the mean shift clusters may meet the partner vehicle's rear criteria. In most cases, this implies the risk that partner vehicles are not being tracked accurately. In such cases (as indicated by the "no" branch from decision 230), the process collects radar points from additional samples until the criteria are met indicating that the partner vehicle has been confidently identified. continue. In some embodiments, if the system has trouble identifying the rear of the partner vehicle, or for other reasons such as the vehicle stopping, after the radar points become too old, or the process restarts. You can optionally discard the radar point later.

In some embodiments, the partner rear identification process continues to run even after the vehicle length has been determined, or is periodically rerun. Keeping the histogram filled has several advantages. Often, the initial length determination is made while the platoon partner is relatively far away (eg, over 100 feet). Once the rear of the partner vehicle is positively identified, the gap controller causes the vehicles to move closer together by narrowing the gap. When vehicles are close together, radar readings are often more accurate than when the vehicles are more than 100 feet apart. Furthermore, considering that in some situations GPS measurements can be relatively far apart for gap control purposes, more measurements give a better statistical indication of the vehicle's relative position. right. By continuing to run the partner rear identification process, these better measurements can be used to more accurately determine the effective length of the partner vehicle, which is highly desirable for control purposes.

FIG. 5D is a plot showing a set of 1700 detected partner vehicle radar point candidates on the same graph shown in FIG. 5A. The 1700 sample points, including the 98 points shown in FIGS. 5A-5C, were obtained by continuing to run the same radar point classification algorithm. Figures 5E and 5F show the histogram and mean shift center, respectively, for the larger data set. Accordingly, FIG. 5E corresponds to FIG. 5B and FIG. 5F corresponds to FIG. 5C. It can be seen that the larger data set appeared to identify small clusters 293 located near the front of the preceding vehicle, effectively excluding some of the smaller clusters identified in the smaller data set.

Keeping the partner's posterior identification process running has other potential uses as well. For example, some trucks can pull the trailer close to the cab when the truck is traveling on a highway. Thus, there are relatively rare situations in which the effective length of a track may change during the course of the platoon. Such changes can be detected automatically by rerunning or continuing the partner rear identification process.

Over time, histograms and/or mean shift clusters also provide a very good indication of the partner vehicle's radar signature. This known signature of the partner vehicle can be used in many different ways as an independent mechanism for verifying that the proper vehicle is being tracked. For example, if GPS data becomes unavailable or communication between vehicles is interrupted for a period of time, the histogram is used as a check to ensure that the correct vehicle is being tracked by the radar unit. be able to. In situations where the rear of the leading truck is not in the radar view of the following vehicle and the rest of the trailer and tractor is in the radar view, the visible portion of the truck is compared to the histogram signature to determine the relative position of the truck. can be determined, which can be used as a measurement for gap control or as part of automatic or semi-automatic control of following vehicles.

In another example, in situations where radar contact is lost, a new histogram can be started at the appropriate time, and the new histogram can be compared to stored histograms indicative of platoon partners. When concordant, the concordance can be good independent evidence that radar contact has been reestablished with a platoon partner. Similarly, as a method of independently verifying that platoon partners are still being tracked, comparing newly created histograms to stored histograms representing platoon partners at various times during platooning. can be done. This can be a good safety check to verify that the radar unit has not been accidentally switched and locked into a car running alongside the platoon partner. The histogram may be saved as the partner vehicle's radar signature, and can also be shared with other trucks that may later attempt to platoon with that vehicle, which aids in the initial identification process. obtain.

Platoon Partner Position Estimation In a platoon situation, maintaining an accurate model of the expected relative position, velocity, and orientation of each vehicle in the platoon is essential if such information is used to provide an accurate estimate of the gaps between platoon partners. It is useful because it is very useful for control. Such models preferably utilize inputs from multiple different sensing systems and, where practical, include at least some redundant information from the different systems. Providing redundant information from different systems is useful as a double check on the integrity of the data received, and in unavoidable circumstances such as when a system is unable to convey accurate information. It also provides a backup mechanism.

By way of example, gaps between vehicles can be determined using a number of different techniques. One common approach involves the use of distances to platoon partners as detected by radar systems. Radar tends to measure distances between vehicles very accurately, but it is important to ensure that the reported distance is actually the distance to the platoon partner and not to another vehicle or feature. be. The partner vehicle may not be within the radar's field of view, or the radar or radar equipment may not be operating as desired for a short period of time. An independent method of determining the distance between platoon partners is to utilize their respective GPS data. Specifically, the distance between vehicles is the difference between the vehicle's respective GPS positions minus the effective length of the preceding vehicle, and the offset distance between the front of the following vehicle and its GPS receiver. should be. Limitations of using GPS data include that the GPS data does not have a clear enough view of the GPS satellites to determine the position or communication link between vehicles where the GPS receiver has been stationary for some time. factors include the fact that it is not always available. GPS data is also fundamentally limited by the fact that although GPS data accuracy is excellent, it is often not as accurate as desired for gap control. Other systems for measuring distance between platoon partners each have their own advantages and limitations.

If the current gap between vehicles is known, the expected gap at the moment can be estimated based on factors such as the vehicle's current position, relative velocity, and yaw rate. A vehicle's respective speed can also be measured, determined, estimated, and/or predicted in a variety of different ways. For example, wheel speed sensors can be used to provide a relatively accurate indication of the current speed of each vehicle. Knowledge of the orientation of the vehicle can be used along with knowledge of the vehicle's speed to determine its speed. A radar unit can be used to measure the relative velocity of the platoon partner. Information about other factors such as torque demand, vehicle weight, engine characteristics, and road grade can be used to predict future vehicle speed.

In the context of radar system control, knowing where the leading vehicle is expected to be relative to the radar unit on the trailing vehicle means that one or more objects detected by the radar unit correspond to the rear of the leading vehicle. It can be very helpful in deciding whether to Thus, in some embodiments, the radar system controller (or other controller whose determination may be utilized by the radar system controller) maintains an estimate of the partner vehicle's current position, heading and relative velocity relative to the radar unit. Contains an estimator. One suitable radar scene processor 600 including a position/state estimator 612 is shown in FIG.

In the embodiment shown in FIG. 6, radar scene processor 600 includes gap monitor 610 and partner identifier 620 . Gap monitor 610 tracks the position of the rear of the partner vehicle based on the radar measurements (after the rear of the partner vehicle has been identified) and sends the corresponding radar position and velocity measurements to the gap controller and/or the rear of the partner vehicle. or configured to report to any other entity interested in such measurements made by the radar unit. One particular implementation of the gap monitoring algorithm is described below with reference to the flowchart of FIG.

In the illustrated embodiment, the gap monitor 610 is used to determine a current estimate of the partner vehicle's position relative to the host vehicle and predict the partner vehicle's expected position when the next radar sample is taken. It includes a position/state estimator 612 having a Kalman filter 615 with a As will be described in more detail with respect to FIG. 7, in the illustrated embodiment, the position/state estimator 612 provides an estimate of the detected radar scene and, e.g., the predicted state (e.g., position, velocity, etc.) of the preceding vehicle. It utilizes both the host and partner vehicle's respective GPS positions in value, wheel speeds, and other available vehicle state information such as inertial measurements. These state estimates can then be used to help interpret the received radar scene. That is, having a reasonable estimate of where the partner vehicle is likely to be within the context of the radar scene allows the gap monitor 600 to extract the partner vehicle from the radar scene, which may include a set of detected objects. It helps to properly identify radar return objects corresponding to the rear. This helps ensure that properly detected points are used in gap control. It also helps identify situations where the tracker is not confident enough about an object (if any) detected by the radar in a particular scene sample that pinpoints the location of the rear of the partner vehicle. As a result, such samples can be disregarded, ignored, or otherwise handled appropriately in the context of the gap control algorithm. One particular Kalman filter design that is well suited for use in position/state estimator 612 is described below with respect to FIG.

Partner identifier 620 includes its own position/state estimator 622 , histogram 624 , clustering algorithm 625 that produces mean shift clusters 626 , and partner length estimator 627 . Partner identifier 620 identifies the rear of the partner vehicle by executing an algorithm such as the algorithm described above with respect to FIG. As part of that process, the histogram 624 is filled. Although the histogram is shown schematically as part of the partner identifier 620, the histogram is simply a data structure that can be physically located in any suitable location, and various It can be used by other processes and controllers. Partner length estimator 624 is configured to determine the length of the partner vehicle (including its anterior and posterior relative to its GPS reference position) based on the histogram and other available information.

The position/state estimator 622 within the partner identifier 620 functions similarly to the position/state estimator 612 described above and may also include a Kalman filter 623 . A significant difference between the position state estimator 622 used for partner identification and the position/state estimator 612 is that it is not known during identification which radar point corresponds to the rear of the partner truck. , the radar unit samples cannot be used as part of the position/state estimate.

Position/state estimation, partner detection, partner length estimation and gap monitoring algorithms may be executed on a radar tracking processor dedicated to radar tracking, or implemented on a processor that performs other gap or platoon management tasks as well. may Each algorithm may be implemented as a separate computing process, or they may be integrated in various ways with each other and/or other functions in various computing processes. Other embodiments may use discrete or programmable logic to implement the functions described. It will be apparent that a wide variety of models can be used to track the position of the partner vehicle's rear relative to the radar unit and estimate future positions. Two particular position/state estimators are shown schematically as part of FIG. 6, and a method that can be used to estimate the current position at any given radar sample time is shown in FIG. is shown in the flow chart of

Referring now to FIG. 7, a method for tracking a partner vehicle and estimating its future position based in part on information received from the radar unit will be described. In the illustrated embodiment, a similar process can be used by the leading vehicle to track the trailing vehicle, or to track parallel vehicles to each other, but the trailing vehicle tracks the location behind the leading vehicle. is doing. The method described assumes that there is a reasonable estimate of the position of the rear of the partner vehicle, which can be initially determined using the method described above with respect to FIG. 2, or in any other suitable manner. can be determined to For example, if the effective length of the vehicle ahead is known, an initial estimate for the relative position of the rear of the vehicle ahead can be estimated based on the GPS position data.

Each time a new radar scene is received (step 502), a determination is made as to whether any radar object points (targets) match the expected position and relative velocity of the rear of the partner vehicle (step 504). This is preferably a probabilistic determination that concludes that the "matching" target most likely actually represents the rear of the partner vehicle. Matching targets provide one method for determining whether or not to quantify the uncertainty associated with the putative location. If the radar target point is within the uncertainty of the expected position, it can be considered a match. As described in more detail below, in some embodiments, Kalman filtering is used to estimate the position of the rear of the partner vehicle and quantify the uncertainty. Kalman filtering is particularly suitable because it inherently adjusts the uncertainty level based on the perceived accuracy of the measurement.

The closest radar identified in the radar scene if two or more of the reported radar target points match the estimated position within the range defined by the uncertainty factor (sometimes called the uncertainty ball). Object points are treated as "matching" targets. In this determination, the "closest" match may be selected based on a combination of metrics including longitudinal position, lateral position, relative velocity, and the like.

If a match is found, the radar tracker sends the distance to the matched object and the relative velocity of the matched object to the gap controller 112 as the current gap and relative velocity with the rear of the partner vehicle (step 506). In some embodiments, the only information transmitted is the longitudinal distance to the rear of the trailer and its relative speed. While currently available radar equipment is generally very good at measuring range and relative velocity, it is difficult to accurately measure lateral velocity or exact lateral position relative to an identified object. It's not very good at providing information. However, if the radar unit used can accurately measure other useful attributes of the target, such as lateral velocity, acceleration, etc., that information can optionally be transmitted as well.

If a match is found, the best matching target is also used to update the radar tracked position and velocity estimates for the rear of the truck (step 508). The position and velocity estimates are then propagated in time to the expected position for the next radar sample at step 510 . That is, the logic estimates the expected location of the rear of the truck when the next radar sample is expected. Since the radar samples are provided at regular intervals, this is a relatively simple problem and it is easy to determine the timing of the next expected sample. For example, if the radar sample rate is 20 Hz, the next sample can be expected to occur 0.05 seconds after the last sample. If the leading and following vehicles are traveling at exactly the same speed and both vehicles are traveling in the same direction, the "expected" position of the rear of the leading vehicle is the last detected position of the leading vehicle's rear. position is exactly the same as However, if one vehicle is turned or turned slightly with respect to the other vehicle, the vehicles are often traveling at slightly different speeds and possibly in slightly different directions. . For example, using a simple example, if a following vehicle is moving in exactly the same direction as the preceding vehicle at a constant speed of 1.00 m/s faster than the preceding vehicle, the rear of the preceding vehicle will be detected by the next radar sample. is expected to be 5 cm closer to the preceding vehicle (0.05 seconds after the last sample was taken). Simple trigonometry can be used to determine the probable position if the vehicles have turned or are turning slightly with respect to each other. Of course, any other quantity of relevant variables known to or obtainable by the radar system controller may also be considered in calculating expected position and velocity to further improve the estimation. can. These may include acceleration of each of the vehicles (measured or estimated), heading and/or turning speed of each of the two vehicles, and the like. Factors that can affect the speed, acceleration, or turning speed of the vehicle, such as each vehicle torque request, current grade, vehicle weight, etc., can also be used to further refine the estimation.

In addition to propagating the position estimate in time, the uncertainty estimate is updated as represented by block 512, as explained in more detail below.

After the position estimate has been propagated in time and the uncertainty estimate has been updated, the process proceeds as depicted in the flow chart of FIG. 7 by returning to step 502 where the next radar scene sample is received. Repeat for the next sample. Propagating the estimated position in time is particularly useful in step 504, where a current estimate of the position of the rear of the preceding vehicle is used to determine whether a match occurs. The current estimate of the position of the preceding vehicle can be expected (and certainly will) to change over time. Subsequently, for each radar sample, the current best estimate of the position of the rear of the leading vehicle can be used, which helps ensure that the partner vehicle is accurately tracked.

As alluded to above, the platoon system preferably utilizes multiple independent or partially independent mechanisms to track the position and velocity of each vehicle. For example, as described above, the platoon controller can access GPS location data that provides an independent mechanism for determining the relative position of the platoon vehicle. The platoon controller also has access to wheel speed data, which provides an alternative mechanism for determining the speed of each, and thus the relative speed of the platoon partners. Such data regarding the host vehicle is available from host vehicle sensors. Partner vehicle data is available via a communication link (eg, DSRC link, cellular link, or other available communication method).

each time a new GPS position estimate is received (as represented by box 520 in FIG. 7), the current GPS position estimate is used to update the radar tracking position and velocity estimates (step 523); The updated position and velocity estimates are propagated in time for the expected reception of the next radar sample, as represented by step 510 . In parallel, each time a new wheel speed estimate is received (as represented by box 530 in FIG. 7), the current wheel speed estimate is used to update the radar tracking position and speed estimates. (step 533 ) and the updated position and velocity estimates are propagated in time to the expected reception of the next radar sample, as represented by step 510 . Similarly, each time a new inertial measurement is received, such as yaw rate, vehicle heading (heading), vehicle pitch and/or vehicle roll (as represented by box 540), the current inertial measurement is used. Radar tracking position and velocity estimates are updated (step 542).

GPS position, wheel speed and inertial measurements are preferably updated relatively quickly, often (but not necessarily) more frequently than radar samples. As an example, a GPS update frequency in the range of 25-500 Hz, such as 50 Hz, has been found to work well for open road platoon control applications. It is not necessary to update the GPS position, wheel speed and/or inertial measurements at the same sample rate as each other or the same sample rate as the radar unit, although similar wheel speed and inertial measurement update frequencies may work well. found.

In the illustrated embodiment, updates from radar units, GPS sensors, wheel speed sensors and inertial measurements are processed asynchronously as they are received. Although not required, this helps the most recent sensor input be utilized in estimating the expected relative position and velocity of the platoon vehicle at the time the next radar unit scene sample is received. . This is in contrast to systems where the wheel speed sensor and GPS sensor information is updated every radar unit sample. Synchronous updates may also work well, but using asynchronous updates tends to improve the accuracy of the estimation because the various sensor inputs can be updated more frequently than the sample rate of the radar unit.

Although different types of measurements need not be synchronized with each other, it is preferable to time-synchronize measurements of the same type on different tracks. That is, since the GPS position measurements on the leading truck are synchronized in time with the GPS position measurements on the trailing truck, the relative positions of the trucks can be determined in time at a particular instant. Similarly, since the wheel speed measurements of the leading truck are synchronized in time with the wheel speed measurements of the following truck, the relative velocity of the trucks can be determined at a particular instant in time. The various inertial measurements are also preferably synchronized with each other.

It should be appreciated that since GPS is used and vehicles communicate with each other via communication links, coordinating the timing of various measurements between vehicles is relatively straightforward. As is known, the GPS system provides highly accurate global timing signals. Therefore, the clocks used by the platoon partners can be synchronized with the GPS signals so that various measurements (e.g. GPS position measurements, wheel speed measurements, inertial measurements, etc.) are specific to each track. can be instructed to occur at the synchronous time of Each measurement is also indicated when the measurement was taken (or when similar sensor measurements are considered unsynchronized between vehicles) so that synchronization of the measurements can be verified. May be accompanied by a timestamp.

Propagation of the estimated position in time is then conducted to the rear of the leading vehicle to determine if any of the received radar sample object points (targets) match the expected position of the partner vehicle's rear. It is particularly useful in step 504, which utilizes a current estimate of the position of the . It should be understood that there may be cases where the radar sample target does not match the expected location at the rear of the partner vehicle, as represented by the "no" branch from decision 504 . In such a case, the radar system controller still propagates the position estimate in time so that the position estimate is updated for the next radar sample based on other information the controller has (step 510). Such other information may then include current estimates and may be further updated based on input from other systems (eg, GPS or wheel speed sensors) as previously described.

There may be operating situations in which one or more measurements are expected to be questionable. For example, if the host vehicle is shaken abnormally hard, such as might occur if the wheels drive over potholes or encounter other irregularities in the road, the radar unit will be shaken accordingly. , radar measurement samples taken at that instant are less likely to be accurate and/or useful to the model. Other sensors such as wheel speed and inertial measurement sensors are similarly less likely to be accurate in such cases. In another example, if the truck ahead brakes aggressively, it is more likely that its trailer will move forward or backward than normal, which again indicates that the radar samples taken during such braking are , suggesting that it is likely not useful for predicting the future position of the rear of the trailer. If the controller detects or is made aware that an event is occurring that renders the readings of any particular sensor questionable, the readings from such sensors are safely ignored in the position estimate. can do. In such situations, we continue to use inputs from other sensors that we believe to be more reliable (if any) to update the position model and position estimates over time for each next sample. can continue to propagate within. The uncertainty associated with position estimation can be expected to increase slightly with each sample that is ignored, which is This has the effect of increasing the variation from the estimated position of the rear of the partner vehicle that is allowed.

The location model described above is relatively simple in that it utilizes a relatively small set of measured inputs. This input consists of (1) the received radar scene, which indicates the relative positions and relative velocities of detected objects, and (2) the measured GPS positions of the platoon partners, which can be used to determine their relative positions. (3) the measured wheel velocities of the platoon partners (which can be used to determine their relative velocities); and (4) the measured yaw rate and heading. In other embodiments, if different or additional types of sensor information are available by the radar controller, the position model can be adapted to make use of any relevant information available in position estimation. For example, if the pitch or roll of the vehicle is available, the position model can incorporate such measurements into the position estimate. Roll can be useful because on trucks GPS antennas tend to be placed above the cab at more than 4 meters above the ground (eg, 14-15 feet). At such heights, even relatively small tilts in the roll direction can significantly vary the reported position of each vehicle. Pitch can also be useful for similar reasons. For example, with a gap between platoons of 15 meters, a difference in pitch of just ±2 degrees can result in a difference of 1 meter in the apparent or detected height of an object. With further distance and/or larger pitch variations these differences are amplified. Since many radar devices used in Platoon systems have a relatively narrow field of view, which is farther than expected from their expected position if pitch is not taken into account, expected targets may not be detected or may be detected. This can result in the subject being ignored. Similarly, if longitudinal and/or angular accelerations are available, the position model can incorporate acceleration measurements into position estimates.

Relative vehicle position and In embodiments where velocity and/or direction can be measured relatively accurately, these measurements can be incorporated into the position model in addition to or instead of GPS, wheel speed and inertial measurements. .

In some embodiments, the position model further refines the predicted position at the next radar sample using inputs such as torque demands, brake signals, and/or other operational information about each platoon partner. can be quite sophisticated.

In the primarily described embodiment, the radar sample object points are compared to the estimated (predicted) position and relative velocity of the rear of the partner vehicle. In other embodiments, more or fewer parameters can be compared to identify matches. For example, in some embodiments, a match (or a mismatch) may be based on matching the expected position of the partner vehicle rather than the position and speed/velocity. If the radar unit can reliably report other information such as acceleration, lateral movement speed, etc., such information can also be compared with the corresponding estimates as part of match identification 504 .

An important advantage of the described approach is that the platoon partner is alerted when the preceding vehicle changes lanes, when an intruder cuts in between the platoon vehicles, or when a temporary obstruction occurs in the radar unit. Relative position and velocity estimation can be reliably continued even when the rear is out of the field of view of the radar unit. With such tracking, the platoon partner's radar identification can be more easily re-established when the rear of the platoon partner returns to the field of view of the radar unit. As those skilled in the art will appreciate, this is not an adaptive cruise control system that simply uses radar to track the distance to the vehicle directly in front of the host vehicle, regardless of who is ahead of it. to differ greatly.

The histograms and/or mean shift clusters described above with respect to FIG. 5 are used as another check to verify that the correct vehicle is being tracked by the radar unit, or that some, but not all, of the tracks are tracked by the radar unit. Note that a reference point can be provided when in the field of view.

A notable feature of the method described with respect to FIG. 7 is that the same algorithm is used to determine the relative position/acceleration of the partner vehicle during the initial radar identification of the partner vehicle as described above with respect to FIG. can be estimated. In that situation, the radar tracker 116/600 would not get a good estimate of the position of the rear of the partner vehicle. Thus, because the target does not match the expected position behind the partner vehicle at decision point 504, the measured position is not reported to the gap controller and the radar unit's measurements are not used to update the position and velocity estimates. . The "no" branch is thereby followed from decision point 504 and steps 506 and 508 are therefore skipped. However, other available sensors, including GPS sensors 131, wheel speed sensors 132 and inertial measurement sensors 134 all provide their respective measurements, which are suitable for use in initial identification of partner vehicles. provide a reasonable estimate of the position of

Kalman Filter The method described with respect to FIG. 7 can be implemented using a variety of techniques. One presently preferred embodiment that works particularly well utilizes a Kalman filter. As used herein, the phrase "Kalman filtering" refers to Linear Quadratic Estimation (LQE), as well as extensions and generalizations of LQE such as extended and unscented Kalman filters designed to work with nonlinear systems. is intended to encompass As will be appreciated by those familiar with Kalman filtering in general, Kalman filtering uses a series of measurements containing noise and other inaccuracies observed over time to reduce a single measurement alone. Generates estimates of unknown variables that tend to be more accurate than those on which they are based. The Kalman filter tracks the estimated state of the system and the variance or uncertainty of the estimation. It is particularly suitable for estimating position, velocity and other state information related to gap control due to the inherent error in some measurements and potential unavailability at the time of some desired measurement samples. ing.

The state variables used in the Kalman filter can vary widely depending on the nature of the model used. One particular state arrangement (X) suitable for use in some of the described embodiments involving a pair of platoon tractor-trailer trucks includes:
(1) the longitudinal position (x) of the center of the rear axle of the front truck relative to the center of the rear axle of the rear truck;
(2) the lateral position (y) of the center of the rear axle of the front truck relative to the center of the rear axle of the rear truck;
(3) Azimuth of the leading truck relative to the azimuth of the trailing truck (χ)
(4) Velocity of preceding vehicle (v ₁ )
(5) Velocity of following vehicle (v ₂ )

This can be expressed mathematically as follows.

The estimated state at the next radar sample (X _k+1 ) is a function of the previous state (X _k ) and the covariance matrix (P _k ), which indicates the level of uncertainty in the measurements. The covariance matrix corresponding to the state array (X) represented above is shown in FIG. As those familiar with Kalman filtering in general will appreciate, the estimated state at the next radar sample (X _k+1 ) is the product of the state transition model (A) and the previous state (X _k ). It is equal to the control input model (B) plus the product of any modeled input (u _k−1 ). This can be expressed mathematically as follows.

One particular control input array (U) is
Includes (1) the yaw rate of the forward vehicle (Ψ1) and (2) the yaw rate of the rearward vehicle (Ψ2).

This can be expressed mathematically as follows.

Note that although specific states and modeled input arrays are illustrated, the specific state and control input variables used in any particular implementation can vary widely based on the nature of the estimation model used. be understood.

Kalman filtering is particularly well suited to making the types of position and velocity estimation useful in the techniques described herein. Although Kalman filtering works particularly well, it should be appreciated that other state/space estimation algorithms, such as particle filtering, can be used in alternative embodiments.

One reason Kalman filtering works so well is that most measurements, including GPS, radar, wheel speed and inertial measurements, are prone to various measurement errors. For example, it is not uncommon for any particular GPS reading to be off by more than 1 meter. The covariance matrix (P _k ) quantifies the statistical variation (error) observed in the measurements and uses that knowledge to improve the quality of position and velocity estimates.

Integration into Other Upward Sensor Data Validation In the embodiment described above, the information received from the partner vehicle regarding the state of the partner vehicle is used by the host to measure the characteristics of the partner vehicle by sensors on the host vehicle. help verify or confirm that the data from is actually representative of the partner vehicle. For example, in some described embodiments, information from a leading vehicle regarding its position, velocity, heading, etc. is used by a radar scene processor on a trailing vehicle to predict the expected position and velocity of the leading vehicle. be. These predictions are then used to help determine which (if any) detected radar objects correspond to preceding vehicles. The status information received from the preceding vehicle can be measured (such as measured wheel speed) or predicted (such as predicted speed), which is more reliable in situations where parameters (such as speed) change. obtain.

It should be appreciated that a wide variety of other information/data received from partner vehicles may additionally or alternatively be used to further aid in such verification. This includes the current torque demand of the partner vehicle, brake status (foundation brakes, retarders, engine brakes, and/or other braking equipment associated with heavy trucks), or other partner vehicle status information such as steering angle. obtain. Information may also include status indications, such as indications that turn signals, hazard lights, taillights or other lights are on. Also, qualitative information about the partner vehicle, such as its radar signature, or its visual appearance (e.g., its color tone, identification markers, or some other feature or characteristic that can be readily identified by one of the controllers on the host vehicle) It can also contain general information. It can also include information about the intended or expected behavior of the preceding vehicle, such as changing lanes, taking the next exit, or turning at the next intersection.

In some circumstances, the host vehicle may require the partner vehicle to take certain actions to assist in such identification. The nature of such requests can vary widely, for example, when a rear truck asks a truck in front to turn on a particular light, switch lanes, accelerate or decelerate to a particular speed, honk its horn, etc. There is

Further, it should be appreciated that additional information regarding partner vehicles may also be obtained from a third vehicle, a larger mesh vehicle, or other external sources of information. For example, a third vehicle running alongside the platoon partner may have measured the partner vehicle's position, speed, and/or other characteristics, and that information can be used as another independent check. In another example, a network operations center (NOC) communicating with both platoon partners may be aware of the intended route and communicate that route, or shorter term directions as appropriate, to the host vehicle. can be done. In other situations, information from a partner vehicle may be sent through an intermediary such as a third party vehicle, NOC, or the like. Any of this type of data can be useful, and some information can be particularly useful in situations where communication between vehicles is temporarily lost.

While only certain embodiments of the invention have been described in detail, it should be understood that the invention can be embodied in many other forms without departing from the spirit or scope of the invention. Although the invention has been described primarily in terms of a pair of truck platoons with a forward-facing radar unit positioned in front of the trailing truck, however, the same concept applies to where the radar unit is located on the vehicle. , and/or any type of vehicle operating in any type of connected vehicle application regardless of the direction(s) interrogated by the radar unit. Thus, for example, a rear-facing radar unit on a leading vehicle can be used to identify and/or track trailing vehicles using radar in substantially the same manner as described. Similarly, if omni-directional radar is used, similar approaches can be used to identify and/or track other vehicles using radar regardless of their position relative to the host vehicle.

As alluded to above, the described radar-based vehicle identification and tracking can be used for units in which independent information about the position and/or velocity of one or more other vehicles is known or that interprets the radar data. It can be used in any type of connected vehicle application that is Thus, for example, the techniques described are particularly well suited for use in transportation systems that include three or more vehicles. The described technique is also very well suited for use in autonomous vehicle traffic applications where knowledge of other specific vehicle intent is considered important. Indeed, this is expected to become an important application of the present invention with the growth of the autonomous vehicle and connected vehicle market.

The present invention has been described primarily in terms of identifying and tracking other vehicles using commercially available radar units designed for use in driving automation systems. Such units are typically designed to analyze received radar energy and identify objects deemed relevant to the radar manufacturer. While the described invention works well with such units, they are less restrictive. Rather, both the vehicle identification and vehicle tracking processes are better suited for use with radar units that filter the response less and report reflected radar signal strengths in a more general way than trying to identify specific objects. very suitable. In particular, the statistical nature of radar return binning and rear-of-vehicle detection makes it very suitable for using radar data provided in other forms, such as intensity/position. Furthermore, the invention is not limited to range finding systems using electromagnetic energy in the frequency range of radar. Rather, the same target vehicle identification and/or tracking technology can be applied to LIDAR utilizing different frequency ranges of electromagnetic energy, acoustic-based ranging (e.g., sonar, ultrasound, etc.), or various time-of-flight-based ranging systems, etc. can be readily used with other electromagnetic energy-based distance measurement techniques. The described techniques may also be used with range finding techniques using cameras or stereo cameras, beacon-based techniques where sensors measure beacon signals transmitted from partner vehicles, and/or other techniques.

In some embodiments, the platoon vehicle may have a mechanism, such as a transponder, suitable for identifying itself to the radar unit. When available, information from such devices can be used to further aid in identifying and tracking platoon partners.

Accordingly, the embodiments are to be considered illustrative and not restrictive, and the invention should not be limited to the details given herein, but rather the scope of the appended claims and their equivalents. can be changed within the range of

U.S. patent application Ser. No. 13/542,622 U.S. patent application Ser. No. 13/542,627 US patent application Ser. No. 14/292,583 U.S. Provisional Patent Application No. 61/505076 U.S. Provisional Patent Application No. 62/249898 U.S. Provisional Patent Application No. 62/343819 U.S. Provisional Patent Application No. 62/377970 International Application No. PCT/US2014/030770 International Application No. PCT/US2016/049143 International Application No. PCT/US2016/060167

Claims (14)

A method of tracking a particular leading vehicle using a distance measuring unit mounted on a trailing vehicle, comprising:
(a) obtaining a current sample comprising a set of object points from the distance measuring unit;
(b) obtaining a current estimate of the state of the preceding vehicle that corresponds to the current sample, has an associated state uncertainty, and does not take into account any information from the current sample; and
(c) determining which object point matches the estimated state of the preceding vehicle within the state uncertainty;
(d) if at least one of the object points matches the estimated state of the preceding vehicle within the range of state uncertainty, then the measured state of the preceding vehicle most closely matches the estimated state of the preceding vehicle; and using the measured state of the preceding vehicle in determining successively subsequent estimates of the state of the preceding vehicle, corresponding to subsequent samples in sequence;
tracking the preceding vehicle by repeating steps (a)-(d) multiple times;
including
If no object point in a particular sample matches the estimated state of the preceding vehicle within the range of state uncertainty, then the state uncertainty for the next estimate of the state of the preceding vehicle in sequence. How to increase sexuality . 2. The method of claim 1 , wherein said estimated state includes a plurality of state parameters, said state parameters including position, velocity and orientation parameters. The current state estimate includes a plurality of state parameters, the state parameters including a position parameter indicative of the position of the leading vehicle relative to the trailing vehicle and a speed parameter indicative of the speed of the leading vehicle relative to the trailing vehicle. , a method according to claim 1 or 2 . further comprising at least partially automatically controlling the trailing vehicle to maintain a desired gap between the leading vehicle and the trailing vehicle, each selected object point being measured by the distance measuring unit; , wherein the associated longitudinal distance serves to maintain a desired gap as the currently measured longitudinal distance from the distance measuring unit to the rear of the preceding vehicle. 4. A method according to any one of claims 1 to 3 , processed by a controller. each sample representing a respective position of said object point;
each current estimate of the state of the preceding vehicle includes a current estimate of the position of the preceding vehicle and has an associated position uncertainty;
5. A method according to any one of claims 1 to 4 , wherein the matching object point selected must correspond to the estimated position of the preceding vehicle within the position uncertainty. each sample also indicating the respective relative velocity of said object point;
each current estimate of the state of the preceding vehicle further comprising a current estimate of the relative speed of the preceding vehicle and having an associated speed uncertainty;
The selected matching object point matches the estimated position of the preceding vehicle within the position uncertainty and the estimated speed of the preceding vehicle within the speed uncertainty. 6. The method of claim 5 , wherein: Further, periodically receiving global positioning satellite system (GNSS) location updates based at least in part on the detected GNSS locations of the preceding and following vehicles;
each time a GNSS position update is received, updating the estimated state and state uncertainty of the preceding vehicle based on such GNSS position update;
7. A method according to any one of claims 1 to 6 , comprising Further, periodically receiving vehicle speed updates based at least in part on sensed wheel speeds of the preceding and following vehicles;
each time a vehicle speed update is received, updating the estimated state and state uncertainty of the preceding vehicle based on such vehicle speed update;
8. A method according to any one of claims 1 to 7 , comprising A method according to any preceding claim, wherein steps (a)-(d) are repeated at a sample rate of at least 10 Hertz. 10. A method according to any one of the preceding claims, wherein Kalman filtering is used to estimate the state of the preceding vehicle and the associated state uncertainty . 11. Claims 1 to 10 , wherein the estimated state of the preceding vehicle comprises an estimated position of the rear of the preceding vehicle, and the selected matching object points are taken as measurements of the relative position of the rear of the preceding vehicle. The method according to any one of . 12. Any one of claims 1 to 11 , wherein a controller on the following vehicle maintains a profile of point clusters representing the preceding vehicle, the selected matching point corresponding to one of the point clusters. The method described in section. 13. A method according to any one of claims 1 to 12 , wherein the preceding and following vehicles are trucks participating in a platoon. 14. A method according to any one of the preceding claims, wherein said range finding unit is a radar unit.

Images