A falling object avoidance control method for an autonomous driving vehicle

Abstract

The present invention provides a falling object avoidance control method for an autonomous driving vehicle, which identifies a cargo box area, constructs a falling object detection space, calculates the falling trajectory of the cargo and the driving trajectory of the vehicle when an object is detected to have invaded the detection space, and controls the vehicle according to a preset avoidance strategy based on the relationship between the falling trajectory of the cargo and the driving trajectory of the vehicle. The present invention enables an autonomous driving vehicle to automatically avoid falling objects from a truck, thereby improving driving safety.

Description

Technical Field

[0001] The present invention relates to the field of autonomous driving technology, and in particular to a falling object avoidance control method for an autonomous driving vehicle. Background Art

[0002] When a truck without a cap on the top of the cargo box is driving at high speed, the cargo is more likely to fall after it exceeds the top of the cargo box. The directions of cargo falling on the truck include sideways and backward. The falling cargo has strong kinetic energy, and if you are hit by the falling cargo, it will cause a serious traffic accident.

[0003] With the continuous advancement of autonomous driving technology, human intervention is decreasing, and it is impossible to avoid the cargo falling from the truck in front by the driver's observation. Therefore, there is an urgent need for a falling object avoidance method suitable for autonomous driving vehicles to improve the safety of autonomous driving vehicles. Summary of the invention

[0004] The present invention provides a falling object avoidance control method for an autonomous driving vehicle, aiming to solve the defects in the prior art, realize automatic avoidance of falling objects from trucks by the autonomous driving vehicle, and improve driving safety.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is:

[0006] The present invention provides a falling object avoidance control method for an autonomous driving vehicle, comprising:

[0007] Step 1: Identify the vehicle in front of the vehicle, and proceed to the next step if it is determined that there is a truck;

[0008] Step 2: Identify the cargo box area and mark the cargo box area with a preset target frame;

[0009] Step 3: construct a falling object detection space, wherein the detection space includes a first detection space and a second detection space;

[0010] Step 4: Detect whether there is an object invading the first detection space, if yes, proceed to the next step, otherwise repeat this step;

[0011] Step 5: Detect whether there is an object invading the second detection space. If yes, determine that there is cargo falling and proceed to the next step, otherwise return to step 4;

[0012] Step 6: Calculate the falling trajectory of the goods and the driving trajectory of the vehicle;

[0013] Step 7, determine whether there is an intersection between the falling trajectory of the goods and the driving trajectory of the vehicle, if yes, proceed to step 9, otherwise proceed to the next step;

[0014] Step 8: Calculate the landing point of the cargo according to the falling trajectory of the cargo, and determine whether the landing point of the cargo is on the driving path of the vehicle. If yes, proceed to the next step, otherwise return to step 4;

[0015] Step 9: Control the vehicle according to the preset avoidance strategy.

[0016] Specifically, the calculation of the falling trajectory of the goods includes:

[0017] With the center of the front of the vehicle as the origin, the vehicle's forward direction as the positive x-axis, the vehicle's lateral left direction as the positive y-axis, and the vertical upward direction as the positive z-axis, a coordinate system is established to determine the trajectory equation of the cargo falling:

[0018]

[0019] Where t is the movement time of the falling cargo, and the timing starts when the cargo starts to fall; v x is the x-speed of the cargo when it falls, which is equivalent to the speed of the truck; v y is the y-speed of the cargo when it falls; D is the x-distance between the cargo and the vehicle when it falls; H is the vertical height of the cargo from the vehicle when it falls; g is the acceleration due to gravity.

[0020] Specifically, the step 3 includes:

[0021] Step 301: Establish at least one detection line group on the adjacent surface between the vehicle and the target frame, the detection line group including a first detection line and a second detection line, the first detection line and the second detection line horizontally cross the adjacent surface, the first detection line is set below the top of the cargo box, the second detection line is set below the first detection line, the distance between the first detection line and the top of the cargo box is a first pixel, the distance between the first detection line and the second detection line is a second pixel, the length of the first detection line and the second detection line is the same as the width of the lane where the truck is located, and the vertical projections of the endpoints at both ends are located on the lane line of the lane where the truck is located;

[0022] Step 302: The first detection line and the extension line of the top contour line of the cargo box constitute a first detection space, and the first detection line and the second detection line constitute a second detection space.

[0023] Specifically, at least one of detecting whether an object invades the first detection space in step 4 and detecting whether an object invades the second detection space in step 5 includes:

[0024] Step A, using high frame rate mode to obtain cargo box images;

[0025] Step B, detecting whether the target object appears in the cargo box image, if yes, proceed to the next step, otherwise repeat this step;

[0026] Step C: determine whether the target object invades the detection space.

[0027] Specifically, the step B includes: comparing the change value between two frames of the cargo box image by a preset method to detect whether the target object appears in the cargo box image.

[0028] Specifically, the preset methods include background subtraction method, frame difference method, optical flow method, etc.

[0029] Specifically, the step C comprises:

[0030] Step C1, identifying the real-time coordinates of the highest point and the real-time coordinates of the lowest point of the target object;

[0031] Step C2, calculating the first equation and the second equation;

[0032] Step C3, calculating the cross product result of the first equation and the second equation;

[0033] Step C4: if the cross product result is less than 0, it is determined that the target object has invaded the detection space; otherwise, it is determined that the target object has not invaded the detection space;

[0034] The first equation is: K p1 =[(x 2n-1 -x 2n ) / (y 2n-1 -y 2n )]-[(x p1 -x 2n ) / (y p1 -y 2n )],

[0035] The second equation is: K p2 =[(x 2n-1 -x 2n ) / (y 2n-1 -y 2n )]-[(x p2 -x 2n ) / (y p2 -y 2n )].

[0036] Among them, (x 2n-1 ,y 2n-1 )Detection line L n The left endpoint coordinate value, (x 2n ,y 2n ) indicates the detection line L n The coordinate value of the right endpoint, (x p1 ,y p1 ) represents the real-time coordinates of the highest point of the target object, (x p2 ,yp2 ) represents the real-time coordinates of the lowest point of the target object.

[0037] Specifically, the step 9 includes:

[0038] Step 901: monitor in real time and determine whether the adjacent lane meets the normal lane change conditions. If yes, control the vehicle to change lanes, otherwise proceed to the next step;

[0039] Step 902: determine whether the vehicle will collide with the falling object when decelerating at the maximum acceleration. If so, proceed to the next step. Otherwise, control the vehicle to decelerate at the maximum acceleration.

[0040] Step 903, calculating the collision speed of the vehicle when braking at the maximum acceleration and colliding with the cargo, and determining whether the collision speed is less than the maximum collision safety speed of the vehicle, if so, controlling the vehicle to decelerate at the maximum acceleration, otherwise executing the next step;

[0041] Step 904, control the vehicle's hazard lights and turn signals to turn on, monitor whether the vehicle behind the vehicle is decelerating, if so, return to step 901, otherwise control the vehicle to decelerate at the maximum acceleration.

[0042] Specifically, the collision speed

[0043] Among them, v1 is the speed of the vehicle before braking; a m is the maximum acceleration during braking; v x is the x-speed of the cargo when it falls; D is the x-distance between the cargo and the vehicle when it falls; H is the vertical height of the cargo from the vehicle when it falls; g is the acceleration due to gravity.

[0044] The beneficial effects of the present invention are as follows: the present invention identifies the cargo box area, constructs a falling object detection space, and when an object is detected to have invaded the detection space, calculates the falling trajectory of the cargo and the driving trajectory of the vehicle, and controls the vehicle according to a preset avoidance strategy based on the relationship between the falling trajectory of the cargo and the driving trajectory of the vehicle, thereby realizing automatic avoidance of falling objects from trucks by autonomous driving vehicles and improving driving safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 is a flow chart of a falling object avoidance control method for an autonomous driving vehicle of the present invention;

[0046] Figure 2 It is a schematic diagram of the detection line group and the detection space of the present invention. DETAILED DESCRIPTION

[0047] The following specifically describes the implementation modes of the present invention in conjunction with the accompanying drawings, which are only used for reference and illustration and do not constitute a limitation on the scope of patent protection of the present invention.

[0048] Example 1

[0049] like Figure 1 As shown, this embodiment provides a falling object avoidance control method for an autonomous driving vehicle, comprising:

[0050] Step 1: Identify the vehicle in front of the vehicle, and proceed to the next step if it is determined that there is a truck.

[0051] Step 2: Identify the cargo box area and mark the cargo box area with a preset target frame.

[0052] For example, Figure 2 As shown, when the cargo box is identified, the cargo box area is marked with a green frame.

[0053] Step 3: construct a falling object detection space, wherein the detection space includes a first detection space and a second detection space.

[0054] Step 4: Detect whether there is an object invading the first detection space. If yes, proceed to the next step; otherwise, repeat this step.

[0055] Step 5: Detect whether there is any object invading the second detection space. If yes, determine that there is cargo falling and proceed to the next step, otherwise return to step 4.

[0056] Step 6: Calculate the falling trajectory of the cargo and the driving trajectory of the vehicle.

[0057] In this embodiment, calculating the falling trajectory of the goods includes:

[0058] With the center of the front of the vehicle as the origin, the vehicle's forward direction as the positive x-axis, the vehicle's lateral left direction as the positive y-axis, and the vertical upward direction as the positive z-axis, a coordinate system is established to determine the trajectory equation of the cargo falling:

[0059]

[0060] Where t is the movement time of the falling cargo, and the timing starts when the cargo starts to fall; v x is the x-speed of the cargo when it falls, which is equivalent to the speed of the truck; v y is the y-speed of the cargo when it falls; D is the x-distance between the cargo and the vehicle when it falls; H is the vertical height of the cargo from the vehicle when it falls; g is the acceleration due to gravity.

[0061] The driving trajectory of the vehicle is determined by the vehicle's driving speed and body parameters (such as vehicle length, width, and height).

[0062] Step 7: Determine whether there is an intersection between the falling trajectory of the goods and the driving trajectory of the vehicle. If yes, proceed to step 9; otherwise, proceed to the next step.

[0063] Step 8: Calculate the landing point of the cargo according to the falling trajectory of the cargo, and determine whether the landing point of the cargo is on the driving path of the vehicle. If yes, proceed to the next step, otherwise return to step 4.

[0064] Through the above-mentioned trajectory equation of the falling goods, it is easy to calculate the coordinates of the landing point of the goods, so as to determine whether the landing point is on the driving path of the vehicle.

[0065] Step 9: Control the vehicle according to the preset avoidance strategy.

[0066] Example 2

[0067] Different from Example 1, this example provides a specific implementation method of step 3, including:

[0068] Step 301: Establish at least one detection line group on the adjacent surface between the vehicle and the target frame, the detection line group including a first detection line and a second detection line, the first detection line and the second detection line horizontally cross the adjacent surface, the first detection line is set below the top of the cargo box, the second detection line is set below the first detection line, the distance between the first detection line and the top of the cargo box is a first pixel, the distance between the first detection line and the second detection line is a second pixel, the length of the first detection line and the second detection line is the same as the width of the lane where the truck is located, and the vertical projections of the end points at both ends are located on the lane line of the lane where the truck is located.

[0069] like Figure 2 As shown, the vehicle is traveling behind a truck, and the rear of the cargo box is the side adjacent to the vehicle. Therefore, a group of detection lines L1 and L2 are established across the cross-section of the rear of the cargo box. The detection line L1 is 10 pixels away from the top of the cargo box, and the detection line L2 is 10 pixels below the detection line L1. The lengths of the detection lines L1 and L2 should be greater than the width of the cargo box, for example, they can be equal to the lane width of the truck.

[0070] Note the detection line L n The left and right endpoints of (n = 1, 2, 3 ... k, k represents the number of detection lines) are denoted as T 2n-1 (x 2n-1 ,y 2n-1 ), T 2n (x 2n ,y 2n For example, the left and right endpoints of the first detection line L1 are T1 (x1, y1) and T2 (x2, y2), and the left and right endpoints of the second detection line L2 are T3 (x3, y3) and T4 (x4, y4).

[0071] If the vehicle is traveling on the side of a truck, it is only necessary to establish a detection line group on the side of the cargo box adjacent to the vehicle according to the same method as described above, without establishing a detection line group on the rear of the cargo box.

[0072] Step 302: The first detection line and the extension line of the top contour line of the cargo box constitute a first detection space, and the first detection line and the second detection line constitute a second detection space.

[0073] like Figure 2 As shown, the first detection line L1 and the extension line of the top contour line of the cargo box constitute the first detection space J1, and the first detection line L1 and the second detection line L2 constitute the second detection space J2.

[0074] Example 3

[0075] Different from Embodiment 1, this embodiment provides a specific implementation method of at least one of detecting whether an object invades the first detection space in step 4 and detecting whether an object invades the second detection space in step 5, including:

[0076] Step A: Use high frame rate mode to obtain cargo box images.

[0077] In this embodiment, the high frame rate mode is a shooting mode with a frame rate not less than 90 FPS.

[0078] Step B: Detect whether the target object appears in the cargo box image, if yes, proceed to the next step, otherwise repeat this step.

[0079] If no object falls, the cargo box image (background image) will not change, so the target object in this proposal refers to the object other than the cargo box (background).

[0080] In this embodiment, the step B includes: comparing the change value between two frames of the cargo box image by a preset method to detect whether the target object appears in the cargo box image.

[0081] The preset methods include background subtraction method, frame difference method, optical flow method, etc.

[0082] Step C: determine whether the target object invades the detection space.

[0083] In this embodiment, step C includes:

[0084] Step C1: Identify the real-time coordinates (x p1 ,y p1 ) and the real-time coordinates of the lowest point P2 (x p2 ,y p2 ).

[0085] Step C2, calculate the first equation and the second equation.

[0086] In this embodiment, the first equation is: K p1 =[(x2n-1 -x 2n ) / (y 2n-1 -y 2n )]-[(x p1 -x 2n ) / (y p1 -y 2n )],

[0087] The second equation is: K p2 =[(x 2n-1 -x 2n ) / (y 2n-1 -y 2n )]-[(x p2 -x 2n ) / (y p2 -y 2n )].

[0088] Among them, (x 2n-1 ,y 2n-1 )Detection line L n The left endpoint coordinate value, (x 2n ,y 2n ) indicates the detection line L n The coordinate value of the right endpoint, (x p1 ,y p1 ) represents the real-time coordinates of the highest point of the target object, (x p2 ,y p2 ) represents the real-time coordinates of the lowest point of the target object.

[0089] Step C3: Calculate the cross product of the first equation and the second equation.

[0090] That is, calculate K = K p1 ×K p2 .

[0091] Step C4: If the cross product result is less than 0, it is determined that the target object has invaded the detection space; otherwise, it is determined that the target object has not invaded the detection space.

[0092] Example 4

[0093] Different from Example 1, this example provides a specific implementation method of step 9, including:

[0094] Step 901: monitor in real time and determine whether the adjacent lanes meet normal lane change conditions. If yes, control the vehicle to change lanes, otherwise proceed to the next step.

[0095] Step 902: Determine whether the vehicle will collide with falling objects when decelerating at the maximum acceleration. If so, proceed to the next step. Otherwise, control the vehicle to decelerate at the maximum acceleration.

[0096] Step 903, calculating the collision speed of the vehicle when braking at the maximum acceleration and colliding with the cargo, and determining whether the collision speed is less than the maximum collision safety speed of the vehicle. If so, controlling the vehicle to decelerate at the maximum acceleration, otherwise executing the next step.

[0097] In this embodiment, the collision speed

[0098] Among them, v1 is the speed of the vehicle before braking; a m is the maximum acceleration during braking; v x is the x-speed of the cargo when it falls; D is the x-distance between the cargo and the vehicle when it falls; H is the vertical height of the cargo from the vehicle when it falls; g is the acceleration due to gravity.

[0099] The maximum collision safety speed can be obtained from the vehicle manufacturer based on the specific model of the vehicle.

[0100] Step 904, control the vehicle's hazard lights and turn signals to turn on, monitor whether the vehicle behind the vehicle is decelerating, if so, return to step 901, otherwise control the vehicle to decelerate at the maximum acceleration.

[0101] The above disclosure is only a preferred embodiment of the present invention, which cannot be used to limit the scope of protection of the present invention. Therefore, equivalent changes made according to the scope of the patent application of the present invention are still within the scope covered by the present invention.

Claims

1. A falling object avoidance control method for an autonomous driving vehicle, characterized in that: include: Step 1: Identify the vehicle in front of the vehicle, and proceed to the next step if it is determined that there is a truck; Step 2: Identify the cargo box area and mark the cargo box area with a preset target frame; Step 3: construct a falling object detection space, wherein the detection space includes a first detection space and a second detection space; Step 4: Detect whether there is an object invading the first detection space, if yes, proceed to the next step, otherwise repeat this step; Step 5: Detect whether there is an object invading the second detection space. If yes, determine that there is cargo falling and proceed to the next step, otherwise return to step 4; Step 6: Calculate the falling trajectory of the goods and the driving trajectory of the vehicle; Step 7, determine whether there is an intersection between the falling trajectory of the goods and the driving trajectory of the vehicle, if yes, proceed to step 9, otherwise proceed to the next step; Step 8: Calculate the landing point of the cargo according to the falling trajectory of the cargo, and determine whether the landing point of the cargo is on the driving path of the vehicle. If yes, proceed to the next step, otherwise return to step 4; Step 9: Control the vehicle according to the preset avoidance strategy; The step 3 comprises: Step 301: Establish at least one detection line group on the adjacent surface between the vehicle and the target frame, the detection line group including a first detection line and a second detection line, the first detection line and the second detection line horizontally cross the adjacent surface, the first detection line is set below the top of the cargo box, the second detection line is set below the first detection line, the distance between the first detection line and the top of the cargo box is a first pixel, the distance between the first detection line and the second detection line is a second pixel, the length of the first detection line and the second detection line is the same as the width of the lane where the truck is located, and the vertical projections of the endpoints at both ends are located on the lane line of the lane where the truck is located; Step 302: The first detection line and the extension line of the top contour line of the cargo box constitute a first detection space, and the first detection line and the second detection line constitute a second detection space; At least one of detecting whether an object invades the first detection space in step 4 and detecting whether an object invades the second detection space in step 5 comprises: Step A, using high frame rate mode to obtain cargo box images; Step B, detecting whether the target object appears in the cargo box image, if yes, proceed to the next step, otherwise repeat this step; Step C, determining whether the target object invades the detection space; The step C comprises: Step C1, identifying the real-time coordinates of the highest point and the real-time coordinates of the lowest point of the target object; Step C2, calculating the first equation and the second equation; Step C3, calculating the cross product result of the first equation and the second equation; Step C4: if the cross product result is less than 0, it is determined that the target object has invaded the detection space; otherwise, it is determined that the target object has not invaded the detection space; The first equation is: K p1 =[(x 2n-1 -x 2n ) / (y 2n-1 -y 2n )]-[(x p1 -x 2n ) / (y p1 -y 2n )], The second equation is: K p2 =[(x 2n-1 -x 2n ) / (y 2n-1 -y 2n )]-[(x p2 -x 2n ) / (y p2 -y 2n )], Among them, (x 2n-1 ,y 2n-1 )Detection line L n The left endpoint coordinate value, (x 2n ,y 2n ) indicates the detection line L n The coordinate value of the right endpoint, (x p1 ,y p1 ) represents the real-time coordinates of the highest point of the target object, (x p2 ,y p2 ) represents the real-time coordinates of the lowest point of the target object.

2. The falling object avoidance control method for an autonomous driving vehicle according to claim 1, characterized in that: The step B comprises: comparing the change value between two frames of the cargo box image by a preset method to detect whether the target object appears in the cargo box image.

3. The falling object avoidance control method for an autonomous driving vehicle according to claim 2, characterized in that: The preset method includes a background subtraction method, a frame difference method or an optical flow method.

4. The falling object avoidance control method for an autonomous driving vehicle according to claim 1, characterized in that: The calculation of the falling trajectory of the goods includes: With the center of the front of the vehicle as the origin, the vehicle's forward direction as the positive x-axis, the vehicle's lateral left direction as the positive y-axis, and the vertical upward direction as the positive z-axis, a coordinate system is established to determine the trajectory equation of the cargo falling: Among them, t is the movement time of the falling goods, and the timing starts when the goods start to fall; is the x-speed of the cargo when it falls, which is equal to the speed of the truck; is the y-speed of the cargo when it falls; D is the x-distance between the cargo and the vehicle when it falls; H is the vertical height of the cargo from the vehicle when it falls; g is the acceleration due to gravity.

5. The falling object avoidance control method for an autonomous driving vehicle according to claim 4, characterized in that: The step 9 comprises: Step 901: monitor in real time and determine whether the adjacent lane meets the normal lane change conditions. If yes, control the vehicle to change lanes, otherwise proceed to the next step; Step 902: determine whether the vehicle will collide with the falling object when decelerating at the maximum acceleration. If so, proceed to the next step. Otherwise, control the vehicle to decelerate at the maximum acceleration. Step 903, calculating the collision speed of the vehicle when braking at the maximum acceleration and colliding with the cargo, and determining whether the collision speed is less than the maximum collision safety speed of the vehicle, if so, controlling the vehicle to decelerate at the maximum acceleration, otherwise executing the next step; Step 904, control the vehicle's hazard lights and turn signals to turn on, monitor whether the vehicle behind the vehicle is decelerating, if so, return to step 901, otherwise control the vehicle to decelerate at the maximum acceleration.

6. The falling object avoidance control method for an autonomous driving vehicle according to claim 5, characterized in that: The collision speed Vc , in, is the speed of the vehicle before braking; a m is the maximum acceleration during braking; is the x-speed of the cargo when it falls; D is the x-distance between the cargo and the vehicle when it falls; H is the vertical height of the cargo from the vehicle when it falls; g is the acceleration due to gravity.

Images