A path cost analysis method and apparatus based on anti-collision and social constraint conditions

Abstract

Embodiments of the present application relate to a path cost analysis method and device based on anti-collision and social constraint conditions, the method comprising: receiving a set of planning paths P, a non-human target set D and a human target set H at any time t0; performing motion path prediction according to the non-human target set D and the human target set H to obtain corresponding non-human path set G D and human path set G H ; performing cost analysis according to an anti-collision constraint condition that the robot should not collide with any dynamic or static target, a social constraint condition that the robot should maintain a reasonable distance from humans and maximize the social distance cost when it is impossible to maintain a reasonable distance, and a social constraint condition that the robot should not pass through a crowd; performing total cost analysis; and judging whether the trajectory passes based on the anti-collision constraint cost and the total cost. The present application can simultaneously consider anti-collision and social constraint conditions.

Description

Technical Field

[0001] This invention relates to the field of embodied intelligence technology, and in particular to a path cost analysis method and apparatus based on collision avoidance and social constraints. Background Technology

[0002] Delivery robots used in crowded places such as hotels, waiting rooms, and shopping malls often have path planning schemes that only focus on collision avoidance constraints between the robot and dynamic or static obstacles, without considering incorporating social constraints that conform to human living habits. As a result, it is common to see delivery robots passing by at close range or walking through groups of people who are chatting.

[0003] In light of this, we provide a trajectory evaluation method for the trajectory planning module of a robot embodied intelligence system that simultaneously considers collision avoidance and social constraints: collision avoidance, social distancing, and no-crossing-in-crowd constraints are quantified into calculable normalized cost items, and the total cost is calculated by weighted summation of these three types of costs. This ensures both an absolute safe distance between the robot and moving or static targets and adherence to social behavior norms in public places. How to implement this improved method is the technical problem that this invention aims to solve. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by providing a path cost analysis method, apparatus, electronic device, and computer-readable storage medium based on collision avoidance and social constraints. Upon receiving a planned path P, a set of non-human targets D, and a set of human targets H at any given time, this invention first predicts the corresponding non-human path set G based on the non-human target set D and the human target set H using a uniform motion mode. D and human path set G H Then, based on the collision avoidance constraint that the robot should not collide with any moving or static targets, and according to the non-human path set G... D and human path set G H Cost analysis is performed on the planned path P; and based on the social constraint that the robot should maintain a reasonable distance from humans and maximize the cost of social distance when a reasonable distance cannot be maintained, the cost is calculated according to the human path set G. H Perform a cost analysis on the planned path P; and based on the social constraint that robots should not walk through crowds, determine the cost of the human path set G. H A cost analysis is performed on the planned path P; then, a total cost analysis is conducted based on the three types of costs; finally, the trajectory evaluation is judged based on the collision avoidance constraint cost and the total cost. Based on this invention, both an absolute safe distance between the robot and moving / static targets are ensured, while also adhering to social behavior norms in public places.

[0005] To achieve the above objectives, a first aspect of the present invention provides a path cost analysis method based on collision avoidance and social constraints, the method comprising: At any time t0, a set of planned paths P, a set of non-human targets D, and a set of human targets H are received; the planned path P includes N+1 path points p. i The quantity N is a preset positive integer, 0 ≤ index i ≤ N; the path point p i Including time t i Location s i and speed v i , t i+1 =t0+i×T, where T is the preset sampling interval; the first path point p0 is the real-time state of the robot at time t0, and the second to N+1th path points p 1→N The robot has N planning states from time t+T to t+NT; the non-human target set D includes the target states d of multiple non-human targets. j(0) , 1≤indexj≤N D N D The total number of targets is D; the target state d j(0) Let be the real-time state of the j-th non-human target at time t0, including its position s. j(0) and speed v j(0) If the v j(0) If the vector magnitude is 0, then the current non-human target is a static obstacle. j(0) If the vector magnitude is not 0, then the current non-human target is a dynamic obstacle; the human target set H includes the target states h of multiple human targets. k(0) , 1≤indexk≤N H N H The total number of targets H; the target state h k(0) Let s be the real-time state of the k-th human target at time t0, including its position s. k(0) and speed v k(0) All positions and velocities are 2D coordinates and velocity vectors in the map coordinate system of a high-precision bird's-eye view map of the specified scene; the specified scene includes at least a hotel, waiting room, and shopping mall; the robot type is a delivery robot in the specified scene; Based on the non-human target set D and the human target set H, motion path prediction is performed according to the uniform motion mode to obtain the corresponding non-human path set G. D and human path set G H The non-human path set G D Including N D A non-human path The non-human path Includes N+1 path points The path point Including the time t i ,Location and speed The human path set G H Including N H Personal Path The human path Includes N+1 path points The path point Including the time t i ,Location and speed ; Based on the collision avoidance constraint that the robot should not collide with any moving or static targets, and according to the non-human path set G... D and the human path set G H Cost analysis is performed on the planned path P to obtain cost C1; 0≤C1≤1; Based on the social constraint that robots should maintain a reasonable distance from humans and maximize the cost of social distancing when a reasonable distance cannot be maintained, according to the human path set G... H Cost analysis of the planned path P yields cost C2; 0 ≤ C2 ≤ 1; Based on the social constraint that robots should not walk through crowds, and according to the human path set G... H Cost analysis of the planned path P yields cost C3; 0 ≤ C3 ≤ 1; Based on costs C1, C2, and C3, a total cost analysis is performed to obtain the total cost C. all ; Regarding the cost C1 and the total cost C all If the values ​​are all less than their respective anti-collision cost threshold and total cost threshold, the system will identify the trajectory. If yes, a trajectory evaluation pass message will be generated and fed back; otherwise, a trajectory evaluation fail message will be generated and fed back.

[0006] Preferably, the motion path prediction based on the non-human target set D and the human target set H in the uniform motion mode yields the corresponding non-human path set G. D and human path set G H Specifically, it includes: Step 21, set each of the target states d in the non-human target set D. j(0) The corresponding non-human target is taken as the current target; and the motion proceeds at a constant speed, based on the position s corresponding to the current target. j(0) and the speed v j(0) Calculate the corresponding N+1 positions. and the speed ; and by each of the stated times t iThe corresponding and the speed The corresponding path points are formed ; and the N+1 path points corresponding to the current target The corresponding non-human path ; in, , , ; Step 22, set each of the target states h in the human target set H. k(0) The corresponding human target is taken as the current target; and the motion proceeds at a constant speed, based on the position s corresponding to the current target. k(0) and the speed v k(0) Calculate the corresponding N+1 positions. and the speed ; and by each of the stated times t i The corresponding and the speed The corresponding path points are formed ; and the N+1 path points corresponding to the current target The corresponding human path ; in, , , .

[0007] Preferably, the collision avoidance constraint that the robot should not collide with any moving or static target is based on the non-human path set G. D and the human path set G H Cost analysis is performed on the planned path P to obtain cost C1, specifically including: Set the preset distance l rob And the planned path P, the set of non-human paths G D and the human path set G H All position vectors are substituted into a preset normalized collision cost function to calculate the corresponding cost C1; The normalized collision cost function is as follows: ; It is an L2 norm; The lower the cost C1, the lower the probability of a collision event; conversely, the higher the cost C1, the higher the probability of a collision event.

[0008] Preferably, the social constraint that the robot should maintain a reasonable distance from humans and maximize social distancing costs when a reasonable distance cannot be maintained is based on the human path set G. H Cost analysis is performed on the planned path P to obtain cost C2, specifically including: Step 41, for the human path set G H The system identifies whether the string is empty; if so, it sets the target total number N. H If the value is 0, then for the human path set G... H The total number of targets N H To conduct statistics; Step 42, if the total number of targets N H If the value is 0, then set the cost C2 to 0 and output it. Step 43, if the total number of targets N H If the value is greater than 0, then the preset spacing l will be... social and the planned path P and the human path set G H All position vectors are substituted into a preset normalized social distance cost function to calculate the corresponding cost C2 and output it. Among them, l social >l rob ; The normalized social distancing cost function is: , ; C social Let social distancing cost function, The preset social distancing cost threshold, ; The lower the cost C2, the lower the probability of an event that cannot maintain a reasonable distance; conversely, the higher the cost C2, the higher the probability of an event that cannot maintain a reasonable distance.

[0009] Preferably, the social constraint that robots should not walk through crowds is based on the human path set G. H Cost analysis of the planned path P yields cost C3, specifically including: Step 51, for the human path set G H The system identifies whether the string is empty; if so, it sets the target total number N. H If the value is 0, then for the human path set G... H The total number of targets N H To conduct statistics; Step 52, if the total number of targets N HIf the value is 0 or 1, then set the cost C3 to 0 and proceed to step 56; Step 53, if the total number of targets N H If it is greater than 1, then based on the time t i For (N+1)×N H The path points Perform clustering, and group the time t i The same N H The path points Clustering into one class forms a corresponding cluster set P i ; Step 54, combine the cluster point sets P of each cluster. i As the current point set; and the corresponding position s in the planned path P. i As the current robot location; and using any two of the aforementioned path points within the cluster. The relative positional spacing is not greater than the preset maximum spacing l group Cluster the current point set according to the conditions, and calculate the total number of point clusters obtained. The total number of the point clusters was statistically analyzed. When the total number of path points within each point cluster is greater than 0, it is greater than or equal to 2; if the total number of the point clusters is greater than 0... If the value is 0, then set the corresponding passage marker cr. i The total number of the point clusters is 0; If the value is greater than 0, then the traversal marker cr is initialized. i If the value is 0, perform a traversal of all point clusters in the current point set; during this traversal, use the currently traversed point cluster as the current point cluster; identify the edge points of the current point cluster and count the total number of identified edge points; if the total number of edge points is 2, check whether the distance between the center point of the two edge points of the current point cluster and the current robot position is less than a preset distance l. social If the identification is successful, then the traversal marker cr i Reset to 1 and stop the current traversal; if the total number of edge points is greater than 2, then connect all edge points of the current point cluster sequentially to obtain a corresponding polygonal region, and identify whether the current robot position is within the current polygonal region. If so, then set the traversal marker cr. i Reset to 1 and stop the current iteration; Among them, the center point of the two edge points of the point cluster is the average coordinate of the position coordinates of the two edge points; Step 55, the obtained (N+1) traversal markers cr i The cost C3 is obtained by substituting the input into the preset normalized crowd passage cost function. The normalized crowd passage cost function is as follows: ; Step 56: Output the obtained cost C3 as the result of this cost analysis.

[0010] Preferably, the total cost C is obtained by performing a total cost analysis based on costs C1, C2, and C3. all Specifically, it includes: Substituting costs C1, C2, and C3 into a preset total cost function, the corresponding total cost C is obtained. all And output; The total cost function is as follows: ; , , These are the preset first, second, and third cost coefficients.

[0011] A second aspect of the present invention provides an apparatus for implementing the path cost analysis method based on collision avoidance and social constraints as described in the first aspect above. The apparatus includes: a data receiving module, an environmental target trajectory derivation module, a collision avoidance constraint cost assessment module, a social reasonable spacing cost assessment module, a crowd crossing cost assessment module, a total cost assessment module, and an assessment feedback module. The data receiving module is used to receive a set of planned paths P, non-human target sets D, and human target sets H at any time t0; the planned path P includes N+1 path points p. i The quantity N is a preset positive integer, 0 ≤ index i ≤ N; the path point p i Including time t i Location s i and velocity v i , t i+1 =t0+i×T, where T is the preset sampling interval; the first path point p0 is the real-time state of the robot at time t0, and the second to N+1th path points p 1→N The robot has N planning states from time t+T to t+NT; the non-human target set D includes the target states d of multiple non-human targets. j(0) , 1≤indexj≤N D N D The total number of targets is D; the target state d j(0) Let be the real-time state of the j-th non-human target at time t0, including its position s. j(0) and velocity v j(0) If the v j(0) If the vector magnitude is 0, then the current non-human target is a static obstacle. j(0)If the vector magnitude is not 0, then the current non-human target is a dynamic obstacle; the human target set H includes the target states h of multiple human targets. k(0) , 1≤indexk≤N H N H The total number of targets H; the target state h k(0) Let s be the real-time state of the k-th human target at time t0, including its position s. k(0) and velocity v k(0) All positions and velocities are 2D coordinates and velocity vectors in the map coordinate system of a high-precision bird's-eye view map of the specified scene; the specified scene includes at least a hotel, waiting room, and shopping mall; the robot type is a delivery robot in the specified scene; The environmental target trajectory derivation module is used to predict the motion path according to the non-human target set D and the human target set H in a uniform motion mode, and obtain the corresponding non-human path set G. D and human path set G H The non-human path set G D Including N D A non-human path The non-human path Includes N+1 path points The path point Including the time t i ,Location and speed The human path set G H Including N H Personal Path The human path Includes N+1 path points The path point Including the time t i ,Location and speed ; The collision avoidance constraint cost evaluation module is used to evaluate the non-human path set G based on the collision avoidance constraint condition that the robot should not collide with any moving or static targets. D and the human path set G H Cost analysis is performed on the planned path P to obtain cost C1; 0≤C1≤1; The social distance cost assessment module is used to evaluate the social distance cost based on the social constraints that the robot should maintain a reasonable distance from humans and maximize the social distance cost when a reasonable distance cannot be maintained, according to the human path set G. H Cost analysis of the planned path P yields cost C2; 0 ≤ C2 ≤ 1; The crowd passage cost assessment module is used to evaluate the human path set G based on social constraints that prevent robots from passing through crowds. H Cost analysis of the planned path P yields cost C3; 0 ≤ C3 ≤ 1; The total cost assessment module is used to perform total cost analysis based on costs C1, C2, and C3 to obtain the total cost C. all ; The evaluation feedback module is used to evaluate the cost C1 and the total cost C. all If the values ​​are all less than their respective anti-collision cost threshold and total cost threshold, the system will identify the trajectory. If yes, a trajectory evaluation pass message will be generated and fed back; otherwise, a trajectory evaluation fail message will be generated and fed back.

[0012] A third aspect of the present invention provides an electronic device, including: a memory, a processor, and a transceiver; The processor is used to couple with the memory, read and execute instructions in the memory to implement the steps of the method described in the first aspect above; The transceiver is coupled to the processor, and the processor controls the transceiver to send and receive messages.

[0013] A fourth aspect of the present invention provides a computer-readable storage medium storing computer instructions that, when executed by a computer, cause the computer to perform the instructions described in the first aspect.

[0014] This invention provides a path cost analysis method, apparatus, electronic device, and computer-readable storage medium based on collision avoidance and social constraints. As described above, after receiving a planned path P, a set of non-human targets D, and a set of human targets H at any given time, this invention first predicts the corresponding non-human path set G based on the non-human target set D and the human target set H in a uniform motion mode. D and human path set G H Then, based on the collision avoidance constraint that the robot should not collide with any moving or static targets, and according to the non-human path set G... D and human path set G H Cost analysis is performed on the planned path P; and based on the social constraint that the robot should maintain a reasonable distance from humans and maximize the cost of social distance when a reasonable distance cannot be maintained, the cost is calculated according to the human path set G. H Perform a cost analysis on the planned path P; and based on the social constraint that robots should not walk through crowds, determine the cost of the human path set G. HA cost analysis is performed on the planned path P; then, a total cost analysis is conducted based on the three types of costs; finally, the trajectory evaluation is judged based on the collision avoidance constraint cost and the total cost. Based on this embodiment of the invention, both an absolute safe distance between the robot and moving / static targets is ensured, while also adhering to social behavior norms in public places. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of a path cost analysis method based on collision avoidance and social constraints provided in Embodiment 1 of the present invention; Figure 2 This is a module structure diagram of a path cost analysis device based on collision avoidance and social constraints provided in Embodiment 2 of the present invention; Figure 3 This is a schematic diagram of the structure of an electronic device provided in Embodiment 3 of the present invention. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0017] Embodiment 1 of the present invention provides a path cost analysis method based on collision avoidance and social constraints, such as... Figure 1 The schematic diagram shows a path cost analysis method based on collision avoidance and social constraints provided in Embodiment 1 of the present invention. The method mainly includes the following steps: Step 1: At any time t0, receive a set of planned paths P, a set of non-human targets D, and a set of human targets H.

[0018] Here, the planned path P in this embodiment of the invention is obtained from the trajectory planning module of the robot embodied intelligence system, and the non-human target set D and human target set H are obtained from the trajectory planning module or environmental perception module of the robot embodied intelligence system. The robot type is a delivery robot in a specified scenario. The specified scenario in this embodiment of the invention includes at least a hotel, a waiting room, and a shopping mall. In a typical scenario, the delivery robot only contains a set of wheels for movement, without a robotic arm or an end effector. The end effector refers to a gripper or dexterous hand connected to the end of the robotic arm.

[0019] The planned path P in this embodiment of the invention includes N+1 path points p i The quantity N is a preset positive integer, where 0 ≤ index i ≤ N. Path point p i Including time t iLocation s i and velocity v i , t i+1 =t0+i×T, where T is the preset sampling interval; the first path point p0 is the real-time state of the robot at time t0, and the second to N+1th path points p 1→N This represents the N planning states of the robot from time t+T to t+NT. All positions and velocities in this embodiment of the invention are 2D coordinates and velocity vectors in the map coordinate system of a high-precision bird's-eye view map of the specified scene. Position s i Its essence is the geometric center of the robot at time t. i Two-dimensional ground coordinates. Velocity v i Its essence is that the robot center is at time t i A two-dimensional plane velocity vector parallel to the ground.

[0020] The non-human target set D in this embodiment of the invention includes the target states d of multiple non-human targets. j(0) , 1≤indexj≤N D N D Let d be the total number of targets D. Target state d j(0) Let be the real-time state of the j-th non-human target at time t0, including its position s. j(0) and velocity v j(0) If v j(0) If the vector magnitude is 0, then the current non-human target is a static obstacle, such as a wall, information desk, ticket gate, reception desk, stool, sofa, etc.; if v j(0) If the vector magnitude is not zero, then the current non-human target is a dynamic obstacle, such as luggage being pushed, luggage carts, etc., or manned indoor shuttle buses, luggage carts, delivery vehicles, etc. Position s j(0) Its essence is that the target center of a non-human target is at time t. i Two-dimensional ground coordinates. Velocity v j(0) Its essence is that the target center of a non-human target is at time t. i A two-dimensional plane velocity vector parallel to the ground.

[0021] The human target set H in this embodiment of the invention includes the target states h of multiple human targets. k(0) , 1≤indexk≤N H N H The total number of targets H; target state h k(0) Let s be the real-time state of the k-th human target at time t0, including its position s. k(0) and velocity v k(0) Position s k(0) Its essence is that the target center of human goals is at time t. i Two-dimensional ground coordinates. Velocity v k(0)Its essence is that the target center of human goals is at time t. i A two-dimensional plane velocity vector parallel to the ground.

[0022] Step 2: Based on the non-human target set D and the human target set H, predict the motion path according to the uniform motion mode to obtain the corresponding non-human path set G. D and human path set G H .

[0023] Here, the non-human path set G in this embodiment of the invention D Including N D A non-human path Non-human pathways Includes N+1 path points Path points Including time t i ,Location and speed .

[0024] Human path set G in this embodiment of the invention H Including N H Personal Path Human Path Includes N+1 path points Path points Including time t i ,Location and speed .

[0025] The current step 2 specifically includes: Step 21, set the target states d of each target in the non-human target set D. j(0) The corresponding non-human target is taken as the current target; and the motion proceeds at a constant speed, based on the position s of the current target. j(0) and velocity v j(0) Calculate the corresponding N+1 positions and speed ; and by each time t i Its corresponding and speed Form the corresponding path points And based on the N+1 path points corresponding to the current target. Forming corresponding non-human paths .

[0026] in, , , .

[0027] Step 22, set the target states h of each target in the human target set H.k(0) The corresponding human target is taken as the current target; and the motion proceeds at a constant speed, based on the position s of the current target. k(0) and velocity v k(0) Calculate the corresponding N+1 positions and speed ; and by each time t i Its corresponding and speed Form the corresponding path points And based on the N+1 path points corresponding to the current target. Forming corresponding human paths .

[0028] in, , , .

[0029] Step 3: Based on the collision avoidance constraint that the robot should not collide with any moving or static targets, and according to the non-human path set G... D and human path set G H Cost analysis of the planned path P yields cost C1.

[0030] Here, 0≤C1≤1.

[0031] Specifically, this includes: setting a preset distance l rob And the planned path P, the set of non-human paths G D and human path set G H All position vectors are substituted into the preset normalized collision cost function to calculate the corresponding cost C1.

[0032] Here, the preset distance l in the embodiment of the present invention rob The maximum horizontal radius at which a robot can generate a collision threat when it is in a non-moving state, with its center line perpendicular to the ground as its axis of rotation.

[0033] If the delivery robot does not have a robotic arm and end effector, then this preset distance l rob Under normal circumstances, this is the robot's maximum lateral radius r. a The product of l and a preset coefficient ξ, rob =r a ×ξ, 1≤ξ≤1.5, the specific value of this preset coefficient ξ can be customized according to application requirements. If the delivery robot consists of a robotic arm and an end effector, then this preset distance l rob That is, the robot's body span radius r b The product of l with the preset coefficient ξ rob =r b×ξ;This body's transverse radius r b The definition is: when the robot's robotic arm and end effector are extended horizontally, a perpendicular line is drawn from the farthest point of the end effector to the robot's center line perpendicular to the ground. The length of this perpendicular line segment is the body's transverse radius r. b .

[0034] The normalized collision cost function in this embodiment of the invention is: ; in, This refers to the L2 norm, which is the method for calculating the vector magnitude.

[0035] In this embodiment of the invention, the lower the cost C1, the lower the probability of a collision event; conversely, the higher the cost C1, the higher the probability of a collision event.

[0036] Step 4: Based on the social constraints that robots should maintain a reasonable distance from humans and maximize social distancing costs when a reasonable distance cannot be maintained, and according to the human path set G... H Cost analysis of the planned path P yields cost C2.

[0037] Here, 0 ≤ C2 ≤ 1.

[0038] Specifically, it includes: Step 41, for the human path set G H The system identifies whether the string is empty; if so, it sets the target total number N. H If it is 0; otherwise, then for the human path set G H Total number of targets N H To conduct statistics.

[0039] Step 42, if the total number of targets N H If the value is 0, then set cost C2 to 0 and output it.

[0040] Step 43, if the total number of targets N H If the value is greater than 0, then the preset spacing l will be... social And the planned path P and the human path set G H All position vectors are substituted into the preset normalized social distance cost function to calculate the corresponding cost C2 and output it.

[0041] Here, the preset spacing l in the embodiment of the present invention social For a pre-set distance parameter, l social >l rob , l social The specific value can be dynamically set according to the specific application scenario.

[0042] The normalized social distancing cost function in this embodiment of the invention is: , ; Among them, C social Let social distancing cost function, The preset social distancing cost threshold has the following range: It can be dynamically configured according to specific application scenarios.

[0043] The lower the cost C2 in this embodiment of the invention, the lower the probability of an event that cannot maintain a reasonable spacing; conversely, the higher the cost C2, the higher the probability of an event that cannot maintain a reasonable spacing.

[0044] Step 5: Based on the social constraint that robots should not walk through crowds, and according to the human path set G... H Cost analysis of the planned path P yields cost C3.

[0045] Here, 0 ≤ C3 ≤ 1.

[0046] Specifically, it includes: Step 51, for the human path set G H The system identifies whether the string is empty; if so, it sets the target total number N. H If it is 0; otherwise, then for the human path set G H Total number of targets N H To conduct statistics.

[0047] Step 52, if the total number of targets N H If the value is 0 or 1, then set cost C3 to 0 and proceed to step 56.

[0048] Step 53, if the total number of targets is N H If it is greater than 1, then it is based on time t. i For (N+1)×N H Path points Perform clustering, and group time t i The same N H Path points Clustering into one class forms a corresponding cluster set P i .

[0049] Step 54, combine the various cluster point sets P i As the current point set; and the corresponding position s in the planned path P. i As the current robot location; and using any two path points within the cluster. The relative positional spacing is not greater than the preset maximum spacing l group Cluster the current point set according to the conditions, and calculate the total number of the resulting point clusters. Statistical analysis was conducted to determine the total number of point clusters. When the total number of path points within each point cluster is greater than 0, the total number of path points within each point cluster is greater than or equal to 2; if the total number of point clusters is greater than 0, the total number of path points within each point cluster is greater than or equal to 2. If the value is 0, then set the corresponding passage marker cr. i The total number of point clusters is 0; If the value is greater than 0, then initialize the traversal marker cr. i If the value is 0, perform a traversal of all point clusters in the current point set; during this traversal, use the currently traversed point cluster as the current point cluster; identify the edge points of the current point cluster and count the total number of identified edge points; if the total number of edge points is 2, check whether the distance between the center point of the two edge points of the current point cluster and the current robot position is less than the preset distance l. social If the identification is successful, then the passage marker cr will be applied. i Reset to 1 and stop the current traversal; if the total number of edge points is greater than 2, connect all edge points of the current point cluster sequentially to obtain a corresponding polygonal region, and identify whether the current robot position is within the current polygonal region. If so, mark the traversal with a cr. i Reset to 1 and stop the current iteration.

[0050] Here, the preset maximum spacing l in the embodiment of the present invention group Given a pre-set distance parameter, and satisfying l group >l social , l group The specific value can be set according to the specific application scenario. One way to set it is: l group =2×l social In this embodiment of the invention, the two edge points of the point cluster are actually two human individuals at the current time t. i The two corresponding positions are denoted as and The center point of the two edge points of the point cluster is actually the position of the two edge points. , The center position between them is denoted as s. k1-k2 The central location s k1-k2 Through calculation , The average vector is obtained as follows: .

[0051] Step 55, the obtained (N+1) traversal markers cr i Substitute the values ​​into the preset normalized crowd passage cost function to calculate the corresponding cost C3.

[0052] Here, the normalized crowd passage cost function in this embodiment of the invention is: .

[0053] The lower the cost C3 in this embodiment of the invention, the lower the probability of a pedestrian crossing incident; conversely, the higher the cost C3, the higher the probability of a pedestrian crossing incident.

[0054] Step 56: Output the obtained cost C3 as the result of this cost analysis.

[0055] Step 6: Perform total cost analysis based on costs C1, C2, and C3 to obtain the total cost C. all .

[0056] Specifically, this includes substituting costs C1, C2, and C3 into a preset total cost function to calculate the corresponding total cost C. all And output it.

[0057] Here, the total cost function of this invention embodiment is: ; in, , , The first, second, and third cost coefficients are preset. . , , The specific settings can be adjusted to suit the needs of the scenario.

[0058] Step 7, compare cost C1 and total cost C all If the values ​​are all less than their respective anti-collision cost threshold and total cost threshold, the system will identify the trajectory. If yes, a trajectory evaluation pass message will be generated and fed back; otherwise, a trajectory evaluation fail message will be generated and fed back.

[0059] Here, the collision avoidance cost threshold and total cost threshold in this embodiment of the invention are two pre-set threshold parameters, which can be adaptively configured based on the type switching of a specified scenario. In this embodiment, the information receiving end that receives feedback on whether the trajectory evaluation passes or fails is the trajectory planning module of the robot's embodied intelligence system. The trajectory planning module can use the method of this embodiment as an auxiliary post-hoc evaluation means. For example, the trajectory planning module generates multiple candidate trajectories each time it plans a trajectory, calculates the cost of the daily trajectory based on the method of this embodiment, and passes the trajectory with the lowest cost to the downstream module. The trajectory planning module can also embed the method of this embodiment as an evaluation calculator, evaluation function, or evaluation module into the trajectory optimization process of the planning module. For example, the trajectory planning module optimizes the trajectory by optimizing the four costs C1, C2, C3, and C4 each time it plans a trajectory. all The optimized trajectory is the direction in which some and all of the elements reach their minimum values.

[0060] Additionally, it should be noted that the embodiments of the present invention provide two constraints in the social constraints: 1) The social constraint that the robot should maintain a reasonable distance from humans, i.e., the human-robot distance ≥ the preset distance l. social 2) Social constraint that robots should not walk through crowds. The total cost is generated by merging these two social constraints with traditional collision avoidance constraints using a weighted summation method. However, in addition to this, embodiments of the present invention can design more social constraints to be integrated with traditional collision avoidance constraints based on actual application needs.

[0061] One of the newly added social constraints is that robots should avoid face-to-face movement with humans. This prevents scenarios where robots appear to be approaching humans from a human's perspective, thus avoiding stress and discomfort for the individual. It is implemented by: based on the social constraint that robots should avoid face-to-face movement with humans, and according to the planned path P and the human path set G... H Cost analysis yields cost C4. Specific implementation steps include: Step A1, set the human paths G H Various human pathways As the current individual trajectory, and set a corresponding opposing motion marker F2F for the current individual trajectory. k and F2F k Initialize to 0; and initialize to all positions of the current individual trajectory. Perform one iteration; and during this iteration, change the current position. As the current individual position, the time t corresponding to the current individual position i As the current moment; and the velocity corresponding to the current individual position. The vector direction is used as the current individual's line of sight; and a preset small angle θ is added to the left and right sides of the current individual's line of sight, thus obtaining a viewpoint boundary line on the left and right sides of the current individual's line of sight. This preset small angle θ can be customized based on application requirements; and with the current position as the center and a preset spacing l, social A circular region is obtained by drawing a circle with a preset coefficient β as the radius. This preset coefficient β is greater than or equal to 1, and its specific value can be customized based on application requirements. The fan-shaped area between the left and right view boundary lines of the current circular region is taken as the visual pressure area of ​​the current individual. The position s in the planned path P corresponding to the current moment is also considered. i The current robot position is used as the reference point; it is then determined whether the current robot position is within the visual pressure area; if so, the velocity v corresponding to the current robot position is further set. iThe vector direction is taken as the current robot's motion direction. Ray 1 is drawn from the current individual's position along the current individual's line of sight, and ray 2 is drawn from the current robot's position along the current robot's motion direction. It is determined whether rays 1 and 2 intersect within the visual compression area and whether they overlap. If rays 1 and 2 overlap or intersect within the visual compression area, the motion marker F2F corresponding to the current individual's trajectory is changed. k Reset to 1 and exit the current iteration process.

[0062] Step A2, on all human paths After all iterations are completed, we get N. H motion marker F2F k ; and the obtained N H motion marker F2F k Substitute the values ​​into the preset normalized opposing motion cost function to calculate the corresponding cost C4.

[0063] Here, the normalized reciprocal motion cost function of this embodiment of the invention is: .

[0064] The lower the cost C4 in this embodiment of the invention, the lower the probability of a human-machine face-to-face movement event; conversely, the higher the cost C4, the higher the probability of a human-machine face-to-face movement event.

[0065] After adding cost C4, step 6 above, "based on costs C1, C2, and C3, we perform a total cost analysis to obtain the total cost C," all This will then become step A3 as follows.

[0066] Step A3: Perform a total cost analysis based on costs C1, C2, C3, and C4 to obtain the total cost C. all .

[0067] Here, the total cost function of this invention after adding cost C4 is: ; in, , , , For another set of preset first, second, third, and fourth cost coefficients, the following should be satisfied: coefficient constraints, , , , The specific settings can be adjusted to suit the needs of the scenario.

[0068] Another newly added social constraint is that the robot should move at a low speed in densely populated areas. This can prevent situations where individuals become anxious and try to avoid the robot due to its rapid movement in crowded local areas. It is implemented by: based on the social constraint that the robot should move at a low speed in densely populated areas, and according to the planned path P and the human path set G... H Cost analysis yields cost C5. Specific implementation steps include: Step B1: Use the planned path P as the robot's motion path, and set all positions s along the robot's motion path. i Perform one round of traversal; and during this round of traversal, change the current traversal position s. i The current robot position is set, and the time t corresponding to the current robot position is set. i As the current moment, an overspeed motion marker OV, initialized to 0, is set for the current robot position. i ; and with the current robot position as the center and a preset area radius r loc A circular region with radius G is designated as the current region; and the human path set G is used. H N corresponding to the current time H Location Form a corresponding current crowd point set; and determine the positions of the current crowd points within the current region. The total number is counted to obtain a first total number; the area of ​​the current region is calculated to obtain a first area; the first total number is divided by the first area to obtain the corresponding first density; and it is identified whether the first density is greater than or equal to a preset density threshold; if so, the velocity v corresponding to the current robot position is... i vector magnitude |v i |Identify whether the speed is greater than or equal to a preset speed threshold; if so, mark the overspeed movement corresponding to the current robot position as OV. i Reset to 1.

[0069] Here, the region radius r in this embodiment of the invention loc For a pre-set distance parameter, r loc >l group The preset density threshold in this embodiment is a pre-set density parameter that can be customized according to specific application requirements. Similarly, the preset speed threshold in this embodiment is a pre-set speed scalar parameter that can be customized according to specific application requirements. The unit of the first density in this embodiment is people per square meter.

[0070] Step B2, at all positions s along the robot's motion path i After completing the traversal, the resulting N+1 hypervelocity markers OV will be... iThe cost C5 is calculated by substituting the preset normalized low-speed motion cost function for dense regions.

[0071] Here, the normalized low-speed motion cost function for dense regions in this embodiment of the invention is: .

[0072] The lower the cost C5 of this invention embodiment, the lower the probability of the robot moving too fast in densely populated local areas; conversely, the higher the cost C5, the higher the probability of the robot moving too fast in densely populated local areas.

[0073] After adding cost C5, step 6 in the previous text, "based on costs C1, C2, and C3, we perform a total cost analysis to obtain the total cost C," all "or, in step A3 above, "to obtain the total cost C by performing a total cost analysis based on costs C1, C2, C3, and C4." all This will then become step B3 as follows.

[0074] Step B3: Perform a total cost analysis based on costs C1, C2, C3, and C5 to obtain the total cost C. all Alternatively, the total cost C can be obtained by performing a total cost analysis based on costs C1, C2, C3, C4, and C5. all .

[0075] Here, the total cost function for total cost analysis based on costs C1, C2, C3, and C5 in this embodiment of the invention is as follows: ; in, , , , For another set of preset first, second, third, and fourth cost coefficients, the following should be satisfied: coefficient constraints, , , , The specific settings can be adjusted to suit the needs of the scenario.

[0076] The total cost function for total cost analysis based on costs C1, C2, C3, C4, and C5 in this embodiment of the invention is as follows: ; in, , , , , For another set of preset cost coefficients (first, second, third, fourth, and fifth), the following conditions should be met: coefficient constraints, , , , , The specific settings can be adjusted to suit the needs of the scenario.

[0077] It should also be noted that, in addition to the cost C1 and total cost C given in step 7 above, the embodiments of the present invention are also based on the present invention. all The trajectory evaluation mechanism based on logic can also be adjusted by adding mandatory collision avoidance costs C1 and C5 after the costs C4 and C5 are increased. 2-5 All and some costs constitute a three-level cost subset, and a three-level decreasing trajectory evaluation mechanism is constructed based on this subset. The implementation method is as follows: Trajectory evaluation is performed according to costs C1, C2, C3, C4, and C5 in a decreasing manner across the three-level cost range to obtain corresponding trajectory evaluation pass / fail information and provide feedback. The specific implementation steps include: In step C1, after obtaining costs C1, C2, C3, C4 and C5, the costs C1, C2, C3, C4 and C5 form a first-level cost subset; the costs C1, C2, C3 and C5 form a second-level cost subset; and the costs C1, C2 and C5 form a third-level cost subset.

[0078] Here, the first-level cost subset of this invention must be the complete cost set, including the collision avoidance cost C1 and all social constraint costs; each level cost subset must include the collision avoidance cost C1; while the social constraint costs constituting the second and third-level cost subsets can be customized based on application requirements. Furthermore, this invention will pre-set corresponding cost thresholds for C1, C2, C3, C4, and C5 based on application requirements.

[0079] Step C2: Identify whether all costs in the first-level cost subset meet their respective cost thresholds; if yes, set the corresponding evaluation result as passed and proceed to step C5; if no, proceed to step C3.

[0080] Step C3: Identify whether all costs in the second-level cost subset meet their respective cost thresholds; if yes, set the corresponding evaluation result as passed and proceed to step C5; if no, proceed to step C4.

[0081] Step C4: Identify whether all costs in the third-level cost subset meet their respective cost thresholds; if yes, set the corresponding evaluation result as passed; if no, set the corresponding evaluation result as failed.

[0082] Step C5: Identify whether the obtained evaluation result is passed; if yes, generate trajectory evaluation passed information and provide feedback; if no, generate trajectory evaluation failed information and provide feedback.

[0083] Figure 2 This is a module structure diagram of a path cost analysis device based on collision avoidance and social constraints provided in Embodiment 2 of the present invention. This device can be a terminal device or server implementing the aforementioned method embodiments, or it can be a device that enables the aforementioned terminal device or server to implement the aforementioned method embodiments. For example, the device can be a device or chip system of the aforementioned terminal device or server. Figure 2 As shown, the device includes: a data receiving module 201, an environmental target trajectory derivation module 202, an anti-collision constraint cost assessment module 203, a social reasonable spacing cost assessment module 204, a crowd passage cost assessment module 205, a total cost assessment module 206, and an assessment feedback module 207.

[0084] Data receiving module 201 is used to receive a set of planned paths P, non-human target sets D, and human target sets H at any time t0; the planned path P includes N+1 path points p i The quantity N is a preset positive integer, 0 ≤ index i ≤ N; path point p i Including time t i Location s i and speed v i , t i+1 =t0+i×T, where T is the preset sampling interval; the first path point p0 is the real-time state of the robot at time t0, and the second to N+1th path points p 1→N Let N be the planning states of the robot from time t+T to t+NT; and let D be the set of non-human targets, which includes the target states d of multiple non-human targets. j(0) , 1≤indexj≤N D N D The total number of targets is D; the target state is d. j(0) Let be the real-time state of the j-th non-human target at time t0, including its position s. j(0) and speed v j(0) If v j(0) If the vector magnitude is 0, then the current non-human target is a static obstacle. j(0) If the vector magnitude is not 0, the current non-human target is a dynamic obstacle; the human target set H includes the target states h of multiple human targets. k(0) , 1≤indexk≤N H N H The total number of targets H; target state h k(0) Let s be the real-time state of the k-th human target at time t0, including its position s. k(0) and speed vk(0) All positions and velocities are 2D coordinates and velocity vectors in the map coordinate system of a high-precision bird's-eye view map of the specified scene; the specified scene includes at least a hotel, waiting room, and shopping mall; the robot type is a delivery robot in the specified scene.

[0085] The environmental target trajectory derivation module 202 is used to predict the corresponding non-human path set G based on the non-human target set D and the human target set H in a uniform motion mode. D and human path set G H Non-human path set G D Including N D A non-human path Non-human pathways Includes N+1 path points Path points Including time t i ,Location and speed Human path set G H Including N H Personal Path Human Path Includes N+1 path points Path points Including time t i ,Location and speed .

[0086] The collision avoidance constraint cost assessment module 203 is used to evaluate the collision avoidance constraints that prevent the robot from colliding with any moving or static targets, based on the non-human path set G. D and human path set G H Cost analysis is performed on the planned path P to obtain cost C1; 0≤C1≤1.

[0087] The social distance cost assessment module 204 is used to assess the social constraints of maintaining a reasonable distance between robots and humans, and maximizing the cost of social distance when maintaining a reasonable distance is not possible, based on the human path set G. H Cost analysis of the planned path P yields cost C2; 0 ≤ C2 ≤ 1.

[0088] The crowd passage cost assessment module 205 is used to assess the social constraints that prevent robots from passing through crowds, based on the human path set G. H Cost analysis of the planned path P yields cost C3; 0 ≤ C3 ≤ 1.

[0089] The total cost assessment module 206 is used to perform total cost analysis based on costs C1, C2, and C3 to obtain the total cost C. all .

[0090] The evaluation feedback module 207 is used for the evaluation of cost C1 and total cost C. all If the values ​​are all less than their respective anti-collision cost threshold and total cost threshold, the system will identify the trajectory. If yes, a trajectory evaluation pass message will be generated and fed back; otherwise, a trajectory evaluation fail message will be generated and fed back.

[0091] The path cost analysis device based on collision avoidance and social constraints provided in this embodiment of the invention can execute the method steps in the above method embodiment. Its implementation principle and technical effect are similar, and will not be repeated here.

[0092] It should be noted that the division of the various modules in the above device is merely a logical functional division. In actual implementation, they can be fully or partially integrated into a single physical entity, or they can be physically separated. Furthermore, these modules can be implemented entirely in software via processing element calls; they can be fully implemented in hardware; or some modules can be implemented in software via processing element calls, while others are implemented in hardware. For example, the data receiving module can be a separate processing element, or it can be integrated into a chip in the above device. Alternatively, it can be stored as program code in the memory of the above device, and called and executed by a processing element of the device. The implementation of other modules is similar. Moreover, these modules can be fully or partially integrated together, or they can be implemented independently. The processing element described here can be an integrated circuit with signal processing capabilities. In the implementation process, each step of the above method or each of the above modules can be completed through integrated logic circuits in the hardware of the processor element or through software instructions.

[0093] For example, these modules can be one or more integrated circuits configured to implement the above methods, such as one or more Application Specific Integrated Circuits (ASICs), one or more Digital Signal Processors (DSPs), or one or more Field Programmable Gate Arrays (FPGAs). As another example, when a module is implemented using processing element scheduler code, the processing element can be a general-purpose processor, such as a Central Processing Unit (CPU) or other processor capable of calling program code. Furthermore, these modules can be integrated together as a System-on-a-Chip (SOC).

[0094] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. This computer program product includes one or more computer instructions. When these computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the foregoing method embodiments are generated. The computer described above can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The aforementioned computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the aforementioned computer instructions can be transmitted from one website, computer, server, or data center to another via wired (e.g., coaxial cable, fiber optic, Digital Subscriber Line (DSL)) or wireless (e.g., infrared, wireless, Bluetooth, microwave, etc.) means. The aforementioned computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The aforementioned available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid-state disks (SSDs)).

[0095] Figure 3 This is a schematic diagram of an electronic device provided in Embodiment 3 of the present invention. This electronic device can be a terminal device or server implementing the methods of the aforementioned embodiments, or it can be a terminal device or server connected to the aforementioned terminal device or server implementing the methods of the aforementioned embodiments. Figure 3 As shown, the electronic device may include: a processor 301 (e.g., CPU), a memory 302, and a transceiver 303; the transceiver 303 is coupled to the processor 301, and the processor 301 controls the transmission and reception operations of the transceiver 303. The memory 302 may store various instructions for performing various processing functions and implementing the processing steps described in the foregoing embodiments. Preferably, the electronic device involved in the embodiments of the present invention further includes: a power supply 304, a system bus 305, and a communication port 306. The system bus 305 is used to realize communication connections between components. The communication port 306 is used for communication between the electronic device and other peripherals.

[0096] exist Figure 3The system bus 305 mentioned can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This system bus can be divided into address bus, data bus, control bus, etc. For ease of representation, it is represented by only one thick line in the figure, but this does not indicate that there is only one bus or one type of bus. The communication interface is used to enable communication between the database access device and other devices (e.g., clients, read-write libraries, and read-only libraries). Memory may include Random Access Memory (RAM) and may also include Non-Volatile Memory, such as at least one disk storage device.

[0097] The processors mentioned above can be general-purpose processors, including central processing units (CPUs), network processors (NPs), graphics processing units (GPUs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0098] It should be noted that the embodiments of the present invention also provide a computer-readable storage medium storing instructions that, when run on a computer, cause the computer to perform the methods and processes provided in the above embodiments.

[0099] This invention provides a path cost analysis method, apparatus, electronic device, and computer-readable storage medium based on collision avoidance and social constraints. As described above, after receiving a planned path P, a set of non-human targets D, and a set of human targets H at any given time, this invention first predicts the corresponding non-human path set G based on the non-human target set D and the human target set H in a uniform motion mode. D and human path set G H Then, based on the collision avoidance constraint that the robot should not collide with any moving or static targets, and according to the non-human path set G... D and human path set G H Cost analysis is performed on the planned path P; and based on the social constraint that the robot should maintain a reasonable distance from humans and maximize the cost of social distance when a reasonable distance cannot be maintained, the cost is calculated according to the human path set G. H Perform a cost analysis on the planned path P; and based on the social constraint that robots should not walk through crowds, determine the cost of the human path set G.H A cost analysis is performed on the planned path P; then, a total cost analysis is conducted based on the three types of costs; finally, the trajectory evaluation is judged based on the collision avoidance constraint cost and the total cost. Based on this embodiment of the invention, both an absolute safe distance between the robot and moving / static targets is ensured, while also adhering to social behavior norms in public places.

[0100] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented in hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0101] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A path cost analysis method based on collision avoidance and social constraints, characterized in that, The method includes: At any time t0, a set of planned paths P, a set of non-human targets D, and a set of human targets H are received; the planned path P includes N+1 path points p. i The quantity N is a preset positive integer, 0 ≤ index i ≤ N; the path point p i Including time t i Location s i and velocity v i , t i+1 =t0+i×T, where T is the preset sampling interval; the first path point p0 is the real-time state of the robot at time t0, and the second to N+1th path points p 1→N The robot has N planning states from time t+T to t+NT; the non-human target set D includes the target states d of multiple non-human targets. j(0) , 1≤indexj≤N D N D The total number of targets is D; the target state d j(0) Let be the real-time state of the j-th non-human target at time t0, including its position s. j(0) and velocity v j(0) If the v j(0) If the vector magnitude is 0, then the current non-human target is a static obstacle. j(0) If the vector magnitude is not 0, then the current non-human target is a dynamic obstacle; the human target set H includes the target states h of multiple human targets. k(0) , 1≤indexk≤N H N H The total number of targets H; the target state h k(0) Let s be the real-time state of the k-th human target at time t0, including its position s. k(0) and velocity v k(0) All positions and velocities are 2D coordinates and velocity vectors in the map coordinate system of a high-precision bird's-eye view map of the specified scene; the specified scene includes at least a hotel, waiting room, and shopping mall; the robot type is a delivery robot in the specified scene; Based on the non-human target set D and the human target set H, motion path prediction is performed according to the uniform motion mode to obtain the corresponding non-human path set G. D and human path set G H The non-human path set G D Including N D A non-human path The non-human path Includes N+1 path points The path point Including the time t i ,Location and speed The human path set G H Including N H Personal Path The human path Includes N+1 path points The path point Including the time t i ,Location and speed ; Based on the collision avoidance constraint that the robot should not collide with any moving or static targets, and according to the non-human path set G... D and the human path set G H Cost analysis is performed on the planned path P to obtain cost C1; 0≤C1≤1; Based on the social constraint that robots should maintain a reasonable distance from humans and maximize the cost of social distancing when a reasonable distance cannot be maintained, according to the human path set G... H Cost analysis of the planned path P yields cost C2; 0 ≤ C2 ≤ 1; Based on the social constraint that robots should not walk through crowds, and according to the human path set G... H Cost analysis of the planned path P yields cost C3; 0 ≤ C3 ≤ 1; Based on costs C1, C2, and C3, a total cost analysis is performed to obtain the total cost C. all ; Regarding the cost C1 and the total cost C all If the values ​​are all less than their respective anti-collision cost threshold and total cost threshold, the system will identify the trajectory. If yes, a trajectory evaluation pass message will be generated and fed back; otherwise, a trajectory evaluation fail message will be generated and fed back.

2. The path cost analysis method based on collision avoidance and social constraints according to claim 1, characterized in that, The motion path prediction based on the non-human target set D and the human target set H in the uniform motion mode yields the corresponding non-human path set G. D and human path set G H Specifically, it includes: Step 21, set each of the target states d in the non-human target set D. j(0) The corresponding non-human target is taken as the current target; and the motion proceeds at a constant speed, based on the position s corresponding to the current target. j(0) and the speed v j(0) Calculate the corresponding N+1 positions. and the speed ; and by each of the stated times t i The corresponding and the speed The corresponding path points are formed ; and the N+1 path points corresponding to the current target The corresponding non-human path ; in, ， ， ； Step 22, set each of the target states h in the human target set H. k(0) The corresponding human target is taken as the current target; and the motion proceeds at a constant speed, based on the position s corresponding to the current target. k(0) and the speed v k(0) Calculate the corresponding N+1 positions. and the speed ; and by each of the stated times t i The corresponding and the speed The corresponding path points are formed ; and the N+1 path points corresponding to the current target The corresponding human path ; in, ， ， 。 3. The path cost analysis method based on collision avoidance and social constraints according to claim 1, characterized in that, The collision avoidance constraint that the robot should not collide with any moving or static targets is based on the non-human path set G. D and the human path set G H Cost analysis is performed on the planned path P to obtain cost C1, specifically including: Set the preset distance l rob And the planned path P, the set of non-human paths G D and the human path set G H All position vectors are substituted into a preset normalized collision cost function to calculate the corresponding cost C1; The normalized collision cost function is as follows: ； It is an L2 norm; The lower the cost C1, the lower the probability of a collision event; conversely, the higher the cost C1, the higher the probability of a collision event.

4. The path cost analysis method based on collision avoidance and social constraints according to claim 3, characterized in that, The social constraints, which require robots to maintain a reasonable distance from humans and maximize social distancing costs when a reasonable distance cannot be maintained, are based on the human path set G. H Cost analysis is performed on the planned path P to obtain cost C2, specifically including: Step 41, for the human path set G H The system identifies whether the string is empty; if so, it sets the target total number N. H If the value is 0, then for the human path set G... H The total number of targets N H To conduct statistics; Step 42, if the total number of targets N H If the value is 0, then set the cost C2 to 0 and output it. Step 43, if the total number of targets N H If the value is greater than 0, then the preset spacing l will be... social and the planned path P and the human path set G H All position vectors are substituted into a preset normalized social distance cost function to calculate the corresponding cost C2 and output it. Among them, l social >l rob ; The normalized social distancing cost function is: ， ； C social Let social distancing cost function, The preset social distancing cost threshold, ; The lower the cost C2, the lower the probability of an event that cannot maintain a reasonable distance; conversely, the higher the cost C2, the higher the probability of an event that cannot maintain a reasonable distance.

5. The path cost analysis method based on collision avoidance and social constraints according to claim 1, characterized in that, The social constraint that robots should not walk through crowds is based on the human path set G. H Cost analysis of the planned path P yields cost C3, specifically including: Step 51, for the human path set G H The system identifies whether the string is empty; if so, it sets the target total number N. H If the value is 0, then for the human path set G... H The total number of targets N H To conduct statistics; Step 52, if the total number of targets N H If the value is 0 or 1, then set the cost C3 to 0 and proceed to step 56; Step 53, if the total number of targets N H If it is greater than 1, then based on the time t i For (N+1)×N H The path points Perform clustering, and group the time t i The same N H The path points Clustering into one class forms a corresponding cluster set P i ; Step 54, combine the cluster point sets P of each cluster. i As the current point set; and the corresponding position s in the planned path P. i As the current robot location; and using any two of the aforementioned path points within the cluster. The relative positional spacing is not greater than the preset maximum spacing l group Cluster the current point set according to the conditions, and calculate the total number of point clusters obtained. The total number of the point clusters was statistically analyzed. When the total number of path points within each point cluster is greater than 0, it is greater than or equal to 2; if the total number of the point clusters is greater than 0... If the value is 0, then set the corresponding passage marker cr. i The total number of the point clusters is 0; If the value is greater than 0, then the traversal marker cr is initialized. i If the value is 0, perform a traversal of all point clusters in the current point set; during this traversal, use the currently traversed point cluster as the current point cluster; identify the edge points of the current point cluster and count the total number of identified edge points; if the total number of edge points is 2, check whether the distance between the center point of the two edge points of the current point cluster and the current robot position is less than a preset distance l. social If the identification is successful, then the traversal marker cr i Reset to 1 and stop the current traversal; if the total number of edge points is greater than 2, then connect all edge points of the current point cluster sequentially to obtain a corresponding polygonal region, and identify whether the current robot position is within the current polygonal region. If so, then set the traversal marker cr. i Reset to 1 and stop the current iteration; Among them, the center point of the two edge points of the point cluster is the average coordinate of the position coordinates of the two edge points; Step 55, the obtained (N+1) traversal markers cr i The cost C3 is obtained by substituting the input into the preset normalized crowd passage cost function. The normalized crowd passage cost function is as follows: ； Step 56: Output the obtained cost C3 as the result of this cost analysis.

6. The path cost analysis method based on collision avoidance and social constraints according to claim 1, characterized in that, The total cost C is obtained by performing a total cost analysis based on costs C1, C2, and C3. all Specifically, it includes: Substituting costs C1, C2, and C3 into a preset total cost function, the corresponding total cost C is obtained. all And output; The total cost function is as follows: ； , , These are the preset first, second, and third cost coefficients.

7. An apparatus for performing the path cost analysis method based on collision avoidance and social constraints as described in any one of claims 1-6, characterized in that, The device includes: a data receiving module, an environmental target trajectory derivation module, an anti-collision constraint cost assessment module, a social reasonable spacing cost assessment module, a crowd passage cost assessment module, a total cost assessment module, and an assessment feedback module; The data receiving module is used to receive a set of planned paths P, non-human target sets D, and human target sets H at any time t0; the planned path P includes N+1 path points p. i The quantity N is a preset positive integer, 0 ≤ index i ≤ N; the path point p i Including time t i Location s i and velocity v i , t i+1 =t0+i×T, where T is the preset sampling interval; the first path point p0 is the real-time state of the robot at time t0, and the second to N+1th path points p 1→N The robot has N planning states from time t+T to t+NT; the non-human target set D includes the target states d of multiple non-human targets. j(0) , 1≤indexj≤N D N D The total number of targets is D; the target state d j(0) Let be the real-time state of the j-th non-human target at time t0, including its position s. j(0) and velocity v j(0) If the v j(0) If the vector magnitude is 0, then the current non-human target is a static obstacle. j(0) If the vector magnitude is not 0, then the current non-human target is a dynamic obstacle; the human target set H includes the target states h of multiple human targets. k(0) , 1≤indexk≤N H N H The total number of targets H; the target state h k(0) Let s be the real-time state of the k-th human target at time t0, including its position s. k(0) and velocity v k(0) All positions and velocities are 2D coordinates and velocity vectors in the map coordinate system of a high-precision bird's-eye view map of the specified scene; the specified scene includes at least a hotel, waiting room, and shopping mall; the robot type is a delivery robot in the specified scene; The environmental target trajectory derivation module is used to predict the motion path according to the non-human target set D and the human target set H in a uniform motion mode, and obtain the corresponding non-human path set G. D and human path set G H The non-human path set G D Including N D A non-human path The non-human path Includes N+1 path points The path point Including the time t i ,Location and speed The human path set G H Including N H Personal Path The human path Includes N+1 path points The path point Including the time t i ,Location and speed ; The collision avoidance constraint cost evaluation module is used to evaluate the non-human path set G based on the collision avoidance constraint condition that the robot should not collide with any moving or static targets. D and the human path set G H Cost analysis is performed on the planned path P to obtain cost C1; 0≤C1≤1; The social distance cost assessment module is used to evaluate the social distance cost based on the social constraints that the robot should maintain a reasonable distance from humans and maximize the social distance cost when a reasonable distance cannot be maintained, according to the human path set G. H Cost analysis of the planned path P yields cost C2; 0 ≤ C2 ≤ 1; The crowd passage cost assessment module is used to evaluate the human path set G based on social constraints that prevent robots from passing through crowds. H Cost analysis of the planned path P yields cost C3; 0 ≤ C3 ≤ 1; The total cost assessment module is used to perform total cost analysis based on costs C1, C2, and C3 to obtain the total cost C. all ; The evaluation feedback module is used to evaluate the cost C1 and the total cost C. all If the values ​​are all less than their respective anti-collision cost threshold and total cost threshold, the system will identify the trajectory. If yes, a trajectory evaluation pass message will be generated and fed back; otherwise, a trajectory evaluation fail message will be generated and fed back.

8. An electronic device, characterized in that, include: Memory, processor, and transceiver; The processor is configured to be coupled to the memory, read and execute instructions in the memory to implement the method according to any one of claims 1-6; The transceiver is coupled to the processor, and the processor controls the transceiver to send and receive messages.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that, when executed by a computer, cause the computer to perform the method according to any one of claims 1-6.

Images