A dynamic unscheduled interference one-to-many robot system and method of implementing the same

Abstract

The application discloses a kind of dynamic no scheduling interference's one picks up multi-robot system and its implementation method, belong to the field of agricultural multi-agent collaborative control.System includes picking robot and multiple transport robots carrying weight sensor, positioning attitude detection module, communication module and radar obstacle avoidance module, system also integrates communication and attitude monitoring as auxiliary safety function.The application proposes seamless connection scheduling method based on real-time load monitoring, according to cumulative load and harvesting operation progress Automatic trigger idle transport robot alternation relay;For orchard narrow road, establish multi-machine dynamic interference avoidance method based on predicted collision time, combined with "load, distance, task" three-level dynamic road right priority rules, realize the autonomous wrong car and conflict-free operation between transport robots, completely eliminate the risk of deadlock and collision in opposite driving and cross path.The application significantly improves the efficiency and safety of collaborative operation in complex orchard environment.

Description

Technical Field

[0001] This invention belongs to the field of agricultural multi-agent collaborative control technology, specifically involving a dynamic, non-scheduled, multi-robot harvesting and transportation system and its implementation method. It is particularly suitable for collaborative harvesting operations of economic forest fruits such as mangoes and citrus fruits, where the individual fruits are heavy, the orchard terrain is complex, and the operating roads are narrow. Background Technology

[0002] Large-scale mechanized harvesting in orchards is a core direction for the development of modern smart agriculture. Taking mango orchard harvesting as an example, mangoes are large in weight and have a high density of fruit per tree, making it easy for the fruit collection bins to quickly reach their full capacity. Traditional orchard harvesting robots generally adopt a fixed pairing operation mode of "one harvester and one transporter." When the transport robot is fully loaded and heading to the unloading point, the harvesting robot can only stop and wait, and the fruit cannot be continuously collected. This results in poor continuity of the overall harvesting operation and low overall operational efficiency.

[0003] While existing multi-robot collaborative harvesting solutions can alleviate the downtime problem of harvesting equipment to some extent, they still have four core technical shortcomings in the unstructured operating environment of hilly mango orchards:

[0004] (1) The mining and transportation scheduling logic is static and fixed, resulting in poor harvesting continuity.

[0005] Existing solutions mostly adopt a fixed-point, timed mechanical handover logic, without dynamically matching the real-time operation progress of the picking robot with the load data of the transport robot. After the transport robot is fully loaded, the picking robot has to wait for a long time for the empty vehicle to return; the handover timing cannot adapt to the fruit tree harvesting progress. Switching too early will interrupt the complete harvesting process of a single fruit tree, while switching too late will cause the fruit box to overflow and the fruit to be damaged by bumps.

[0006] (2) The multi-machine conflict avoidance mechanism is simple, and deadlock and collision are prone to occur when passing through narrow passages.

[0007] The roads in the mango orchard are mostly narrow, single-lane passages, making it easy for oncoming vehicles to meet each other and for path intersections to occur when multiple transport robots are shuttling back and forth. Existing solutions rely solely on single-unit LiDAR for local obstacle avoidance, lacking a multi-vehicle collaborative right-of-way arbitration mechanism. This fails to address the global path interference problem among multiple agents, easily leading to malfunctions such as two-way blockages, deadlocks, and vehicle collisions.

[0008] (3) Without considering the terrain and posture for safety constraints, the fully loaded robot has a high risk of tipping over and sliding down slopes.

[0009] Mango orchards are mostly built on hilly and sloping terrain, with uneven roads and many lateral tilting sections. When transport robots are fully loaded with fruit boxes, their center of gravity rises significantly. If only the shortest path is used as the driving planning objective, safety accidents such as wheel slippage, vehicle rollover, and slippage on steep slopes and highly lateral tilting sections are very likely to occur. Current technology does not achieve real-time closed-loop monitoring of vehicle posture and dynamic path replanning.

[0010] (4) The number of transport robots deployed is determined by human experience, resulting in an imbalance in resource allocation.

[0011] The existing system does not have a quantitative configuration model and relies solely on the experience of operators to determine the number of transport vehicles. When the number of vehicles is insufficient, the picking robots will remain idle and waiting. When the number of vehicles is redundant, a large number of transport robots will be idle, resulting in serious waste of hardware costs and equipment energy consumption. It is impossible to dynamically match the optimal number of transport vehicles according to the orchard conditions.

[0012] In summary, current multi-robot systems for orchard harvesting and transportation suffer from shortcomings such as inefficient collaborative scheduling, inability to fundamentally eliminate multi-robot interference, insufficient safety in hillside operations, and lack of quantitative basis for equipment configuration. There is an urgent need for an integrated multi-robot system for harvesting and transportation, along with its control method, that can achieve seamless multi-robot rotation and connection, autonomous intelligent passing in narrow lanes, optimal configuration of the number of transport vehicles, and dynamic anti-rollover protection based on vehicle posture. Summary of the Invention

[0013] This invention addresses the technical shortcomings of existing orchard harvesting and transportation robots, such as low collaborative efficiency, lack of quantitative standards for the number of transport vehicles, susceptibility to deadlock and collision when multiple robots pass through narrow passages, and susceptibility to rollover and slippage when fully loaded on hilly slopes. It provides a dynamic, non-intervention-based one-harvest-multiple-transport robot system and its implementation method, achieving seamless harvesting and transportation relay, autonomous right-of-way negotiation and avoidance among multiple robots, safe driving with adaptive attitude on slopes, and optimal quantitative configuration of the number of transport vehicles.

[0014] The present invention achieves the above-mentioned technical objectives through the following technical means.

[0015] A dynamic, non-interference-based, multi-transport robot system includes one harvesting robot and at least two transport robots with identical configurations. The robots establish bidirectional data transmission via a communication network. Each transport robot is equipped with a walking mechanism, a fruit storage unit, a main control module, and a weight sensor, a positioning and attitude detection module, a radar obstacle avoidance module, and a wireless communication module electrically connected to the main control module. The weight sensor is located at the bottom of the fruit storage unit and is used to collect the cumulative weight data of the fruit inside the storage box in real time. The positioning and attitude detection module integrates a GNSS / INS combined navigation unit and outputs the robot's global coordinates, speed, heading angle, roll angle, and pitch angle in real time. The radar obstacle avoidance module uses lidar to identify obstacles. The wireless communication module is used for synchronous sharing of position, speed, load status, task type, obstacle avoidance point occupancy status, and control commands among the multiple robots.

[0016] In the above technical solution, the fruit temporary storage unit integrates a one-piece fruit box locking device. The locking device has a built-in electromagnetic latch, a self-locking mechanical card holder, and a micro switch. The micro switch is located inside the self-locking mechanical card holder. The electromagnetic latch is used to release a full fruit box or lock an empty box under the control of the main control module. The self-locking mechanical card holder is used to automatically limit and securely lock the fruit box when an empty box is placed in. When the transport robot completes unloading and replaces the empty box, the self-locking mechanical card holder automatically locks the fruit box and presses the micro switch to send an unloading completion electrical signal to the main control module.

[0017] In the above technical solution, the main control module has built-in dynamic connection scheduling, multi-machine dynamic interference avoidance, and slope attitude safety protection;

[0018] The dynamic shuttle scheduling is as follows: based on whether the load data of the weight sensor has reached the preset load threshold, and combined with the harvesting progress of the picking robot, a takeover instruction is dynamically generated, and an idle transport robot is dispatched to the target coordinate for seamless takeover.

[0019] The multi-machine dynamic interference avoidance is as follows: real-time acquisition of the position, driving direction and speed information of surrounding transport robots, calculation and prediction of collision time, and then determination of whether there is an interference risk; when it is determined that there is an interference risk, the passage order of multiple transport robots is determined according to the preset dynamic right-of-way priority rules, and the transport robot with lower priority is driven to drive into the preset avoidance point to park and wait.

[0020] The slope attitude safety protection is as follows: continuously collect the roll angle and pitch angle of the transport robot. If the detected value exceeds the preset safety threshold, the main control module immediately controls the transport robot to perform speed-limited crawling and triggers online path replanning to generate an alternative driving path.

[0021] In the above technical solution, the dynamic shuttle scheduling includes quantitative calculation of the number of transport vehicles and four types of operation judgment scenarios;

[0022] The specific calculation of the number of transport vehicles is as follows: Before operation, the round-trip time of the transport robots is estimated based on the orchard area, fruit tree distribution, and unloading points, and the initial configuration number is set; during operation, when the transport robots are fully loaded and scheduling is triggered, the shortest travel distance is calculated using the current fully loaded point as the starting point and the unloading point as the ending point. Combined with the average travel speed of the transport robot Time spent on a single unloading Calculate the time for a single round trip. The statistics show the round-trip time for all completed trips by the transport robots. This allows for the determination of the average round-trip time. Find the minimum number of transport vehicles required. ,in, This represents the average harvesting efficiency of the harvesting robot. This represents the full load capacity of the transport robots; if the actual number of transport robots configured... Increase the number of transport robots to achieve the desired actual number of transport robots deployed. ;

[0023] The four types of task determination scenarios are as follows:

[0024] (1) The current fruit tree has been harvested and the fruit box has not reached the full load threshold. The picking robot moves to the next target fruit tree and the current transport robot continues to follow and connect.

[0025] (2) The fruit trees have been harvested and the fruit boxes are full. The current transport robot is heading to the unloading point. The main control module sends a scheduling instruction to the transport robot that is in an idle state through the communication network, driving it to the next target fruit tree to wait for pick-up.

[0026] (3) If the fruit trees have not been harvested yet and the fruit boxes are full in advance, the current transport robot will drive to the unloading point. The main control module will use the communication network to dispatch an idle transport robot to the current fruit tree location to take over the operation.

[0027] (4) All transport robots are fully loaded and head to the unloading point. The picking robots pause their harvesting actions until any transport robot completes unloading and resets, after which the harvesting operation resumes.

[0028] In the above technical solution, during the multi-machine dynamic interference avoidance process, the relative distance between two transport robots traveling in opposite directions is calculated using positioning data. Relative speed is calculated by differentiating continuous distance measurements. Solve for the predicted collision time ;like or The assessment indicated a risk of interference, among which, The minimum safe distance threshold is set. This is the minimum safe time threshold.

[0029] In the above technical solution, the dynamic right-of-way priority rules are as follows: the first level is based on load status priority, where a fully loaded transport robot heading to the unloading point has a higher priority than a fully loaded transport robot; the second level is based on distance from the target, where, when the load status is the same, the transport robot closer to its target point has a higher priority; and the third level is based on task type priority, where, when both load and distance conditions are the same, the transport robot heading to the fruit tree transfer point to take over the work has a higher priority than the transport robot returning to the transfer waiting area to wait.

[0030] In the above technical solution, once the priority is determined, the main control module of the transport robot with lower priority automatically retrieves the nearest unoccupied preset avoidance point from the electronic map as the avoidance target, and drives the walking mechanism to enter the avoidance point; the preset avoidance points include road widening sections and intersections, and their occupancy status is synchronized in real time among multiple robots through the communication network.

[0031] In the above technical solution, the system is also equipped with fail-safe protection logic: when the communication interruption time of any transport robot exceeds a preset threshold or the positioning posture detection module signal is lost, the main control module cuts off the driving power of the walking mechanism, the transport robot slows down and parks in place and starts an audible and visual alarm; after the communication or positioning signal is restored, the transport robot returns to the nearest road network node or the waiting area in a low-speed safety mode.

[0032] A method for implementing a dynamic, non-scheduling-interference-free single-collection, multi-transport robot system includes:

[0033] Step 1: The picking robot and multiple transport robots complete power-on self-tests, confirm that each transport robot is equipped with an empty fruit box and parked in the waiting area, and establish a multi-machine real-time location and status sharing communication network.

[0034] Step 2: The picking robot performs the picking operation, and a designated transport robot follows and connects with it; the connection scheduling is dynamically adjusted according to the four types of operation judgment conditions, and the fully loaded transport robot drives to the unloading point, and the idle transport robot seamlessly takes over;

[0035] Step 3: When multiple transport robots are traveling in opposite directions on the same road segment, the main control module calculates the relative distance and predicted collision time of the oncoming vehicles in real time during the robot's travel. If an interference risk is predicted, the low-priority vehicle is automatically driven to the nearest preset avoidance point and parked according to the dynamic right-of-way priority rules, while the high-priority vehicle has priority to pass. When multiple transport robots are traveling in the same direction on the same road segment, the main control module maintains a safe following distance and does not trigger interference avoidance. When a rear transport robot approaches a front transport robot, the front transport robot actively yields at the preset avoidance point.

[0036] Step 4: Collect the vehicle posture of the transport robot in real time during the driving process. When the preset safety threshold is exceeded, the vehicle will crawl at a limited speed and replan to generate an alternative driving path.

[0037] Step 5: After the fully loaded transport robot arrives at the unloading point, it uses the integrated fruit box locking device to replace the fruit box and automatically triggers the unloading completion electrical signal. The main control module then drives the transport robot to re-enter the transfer task queue.

[0038] Step six: Repeat steps two through five until the harvesting task is completed.

[0039] Furthermore, if a vehicle's posture exceeds the limit multiple times on the same road segment, it will be marked as a permanently dangerous road segment, and the route planning will automatically avoid this road segment.

[0040] Compared with the prior art, the present invention has the following technical effects:

[0041] (1) Quantitative scheduling enables seamless harvesting and transportation, significantly reducing downtime during harvesting.

[0042] Relying on real-time load monitoring by weight sensors and equipped with four types of adaptive shuttle scheduling logic, idle transport robots can take over in advance, eliminating long-term downtime waiting.

[0043] (2) Quantitative modeling of the number of transport vehicles to achieve optimal matching of equipment resources.

[0044] A calculation model for the minimum number of transport vehicles is constructed, which unifies and quantifies multiple parameters such as harvesting efficiency, transport distance, load limit, and round-trip time. This eliminates the manual experience-based configuration mode, thereby avoiding harvesting downtime caused by insufficient transport vehicles and preventing energy and cost waste caused by redundant and idle equipment.

[0045] (3) TTC collision prediction and three-level right-of-way arbitration completely solve the deadlock collision problem of multiple vehicles in narrow lanes.

[0046] Predicted collision time (TTC) is introduced to quantify interference risks, and a three-layer right-of-way negotiation mechanism of "load-distance-task" is established, enabling multiple vehicles to autonomously identify avoidance points and pass each other in an orderly manner. Combined with single-unit obstacle avoidance of lidar as a redundant safety guarantee, it solves two types of problems: global path conflict of cooperative multi-unit vehicles and local obstacle avoidance of non-cooperative obstacles.

[0047] (4) Attitude closed-loop monitoring and dynamic path replanning improve the safety level of operations on hilly slopes.

[0048] The system uses GNSS / INS high-frequency data acquisition to measure vehicle roll and pitch angles. When the speed exceeds the limit, it automatically limits the speed and calls the A* algorithm to bypass high-risk road sections. The system has self-learning capabilities and automatically marks and permanently avoids road sections that exceed the limit multiple times. It dynamically distributes the four-wheel drive torque on uphill and downhill slopes, eliminating the risk of rollover and slippage of fully loaded vehicles from three levels: path planning, driving speed, and power distribution. Attached Figure Description

[0049] Figure 1 This is a schematic diagram of the overall architecture and communication connection of the multi-transport robot system of the present invention;

[0050] Figure 2 This is the complete control flowchart of the dynamic shuttle scheduling logic of the present invention (embedded sub-flow for calculating the minimum configuration number of transport vehicles).

[0051] Figure 3 This is a top view of the principle of dynamic interference avoidance and right-of-way passing for multiple vehicles in narrow lanes according to the present invention. Detailed Implementation

[0052] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, but the scope of protection of the present invention is not limited thereto.

[0053] like Figure 1 As shown, a dynamic, non-disruptive, one-to-many robot system includes one harvesting robot and at least two transport robots with the same configuration; all robots establish a fully bidirectional real-time data interaction link through a wireless communication network.

[0054] Each transport robot is equipped with a walking mechanism, a fruit storage unit, and a main control module, as well as a weight sensor, a positioning and attitude detection module, a radar obstacle avoidance module, and a wireless communication module electrically connected to the main control module. The weight sensor is located at the bottom of the fruit storage unit to collect real-time data on the cumulative weight of the fruit inside the storage box. The positioning and attitude detection module integrates a GNSS / INS navigation unit, outputting the robot's global coordinates, speed, heading angle, roll angle, and pitch angle in real time. The radar obstacle avoidance module uses a lidar unit as a redundant local obstacle detection unit to identify non-cooperative obstacles such as fallen branches, rocks, and pedestrians. The wireless communication module is used for synchronous sharing of position, speed, load status, task type, obstacle avoidance point occupancy status, and control information among multiple robots. The system controls the fruit storage unit with an integrated fruit box locking device. This device includes an electromagnetic latch, a self-locking mechanical holder, and a microswitch. The electromagnetic latch, under the control of the main control module, releases a full-load fruit box or locks an empty one. The self-locking mechanical holder automatically limits and securely locks the fruit box when an empty one is placed in the system. The microswitch is located inside the self-locking mechanical holder. After the transport robot finishes unloading and replacing the empty box, the holder automatically locks the fruit box and activates the microswitch. The microswitch sends an unloading completion signal to the main control module. Upon receiving this signal, the main control module automatically drives the walking mechanism back. The main control module incorporates three sets of collaborative control logic: dynamic connection scheduling logic, multi-machine dynamic interference avoidance logic, and slope attitude safety protection logic.

[0055] (A) Dynamic connection scheduling logic: Read the load data of the weight sensor in real time and compare it with the preset full load threshold. It also dynamically generates a relay scheduling instruction based on the harvesting progress of individual fruit trees by the picking robot, and dispatches idle transport robots to the target coordinates to complete a seamless relay; it has a built-in method for quantitatively calculating the minimum configuration of transport vehicles, and dynamically outputs the optimal equipment configuration scale based on the orchard working conditions;

[0056] (B) Multi-machine dynamic interference avoidance logic: The position, heading angle and driving speed of the surrounding transport robots are obtained synchronously through wireless communication, and the predicted collision time (TTC) is calculated to determine the path interference risk. When there is an interference risk, the low-priority robots drive into the nearest empty preset avoidance point to wait in accordance with the three-level dynamic right-of-way priority arbitration passage order, thus eliminating oncoming, crossing and rear-end collision conflicts.

[0057] (C) Slope attitude safety protection logic: continuously collect the roll angle and pitch angle of the transport robot. If the roll angle or pitch angle exceeds the preset safety threshold (i.e., the corresponding road surface tilt or slope is too large), the main control module immediately controls the transport robot to perform speed-limited crawling and starts A* global path online replanning to generate an alternative driving path to avoid high-risk slopes and tilted road sections.

[0058] The system has built-in failure safety protection logic: when the communication interruption time of any transport robot exceeds the preset threshold, or when the positioning posture detection module loses the signal, the main control module cuts off the driving power of the walking mechanism, the transport robot slows down and parks in place and starts the audible and visual alarm; after the communication or positioning signal is restored, the transport robot returns to the nearest road network node or the waiting area in a low-speed safety mode.

[0059] Based on the above-described dynamic, non-scheduling-interference-free single-mining-multiple-transport robot system, this embodiment also provides a method for implementing a dynamic, non-scheduling-interference-free single-mining-multiple-transport robot system, comprising the following steps:

[0060] Step 1: Power-on initialization and network self-test of multiple robots

[0061] The picking robot and all transport robots are powered on and complete hardware self-checks. All transport robots lock empty fruit boxes and stop at the waiting area for docking. The picking robot and transport robot establish a real-time shared wireless communication network for multi-machine status, load an electronic map of the orchard road network, and pre-store the coordinates of preset avoidance points such as road widening and intersections.

[0062] Step 2: Dynamic rotation and shuttle scheduling based on load monitoring

[0063] The picking robot starts the harvesting operation, and a single transport robot is assigned to follow and connect. The replacement scheduling instructions are dynamically issued according to four operating scenarios. Fully loaded vehicles drive to the unloading point, and idle vehicles arrive at the target point in advance to complete the seamless replacement. During the harvesting process, the weight sensor collects the load of the fruit box in real time and calculates the minimum number of transport vehicles.

[0064] Step 3: Narrow-lane multi-vehicle interference avoidance based on TTC and three-level right-of-way priority

[0065] When multiple transport robots travel towards each other on the same road segment, the relative distance between the two vehicles is calculated in real time using positioning data during the transport robot's journey. Calculate relative speed by difference of continuous distance measurement values Solve for the predicted collision time ;like or The assessment indicated a risk of interference, among which, The minimum safe distance threshold is set. The minimum safe time threshold is used; passage permissions are determined according to a three-tier priority system of "load-distance-task". Low-priority vehicles enter and park at the nearest unoccupied yielding point to wait. See [link / reference]. Figure 3When multiple transport robots are traveling in the same direction on the same road segment, the main control module maintains a safe following distance and does not trigger interference avoidance. When a high-speed transport robot behind approaches a low-speed transport robot in front, the transport robot in front actively yields at a preset avoidance point. During this process, the radar obstacle avoidance module directly executes emergency braking when it detects a non-cooperative obstacle. Whether the robots are traveling in opposite directions or in the same direction is determined based on the heading angle.

[0066] Step 4: Slope Adaptive Safe Driving Based on Vehicle Attitude Monitoring

[0067] The positioning and attitude detection module collects the roll angle and pitch angle of the transport robot at high frequency. After the main control module detects that the roll angle or pitch angle exceeds the limit, it limits the speed of the vehicle and crawls, and calls the A* algorithm to replan the detour path. If the vehicle's attitude exceeds the limit multiple times in the same road segment, it is marked as a permanently dangerous road segment, and the global path planning automatically avoids the area. The four-wheel drive or braking torque is dynamically allocated according to the uphill and downhill attitude to prevent slippage and rollover.

[0068] Step 5: Unloading Reset and Task Queue Update

[0069] The fully loaded transport robot arrives at the unloading point. The integrated fruit box locking device completes the unloading of the full box and locks the empty box. The micro switch outputs a signal that the unloading is complete. After receiving the signal, the main control module controls the vehicle to return to the waiting area and enter the idle task queue.

[0070] Step 6: Iterate

[0071] Repeat steps 2 through 5 until the entire orchard harvest is completed.

[0072] like Figure 2 As shown, the dynamic shuttle scheduling in step 2 includes quantitative calculation of the number of transport vehicles and four operational judgment scenarios:

[0073] (1) Calculation process for minimum configuration quantity of transport vehicles:

[0074] Before the operation, the round-trip time of the transport robot is estimated based on the orchard area, fruit tree distribution and unloading point, and the initial configuration quantity is set.

[0075] During operation, when the transport robot is fully loaded and scheduling is triggered, the shortest travel distance is calculated using the current fully loaded point as the starting point and the unloading point as the ending point. Combined with the average travel speed of the transport robot Time spent on a single unloading Then calculate the round trip time:

[0076]

[0077] The total round-trip time of all completed trips by the transport robots is recorded. This allows for the determination of the average round-trip time. Find the minimum number of transport vehicles required.

[0078]

[0079] In the formula, The average harvesting efficiency (kg / h) of the harvesting robot. Full load capacity (kg) for the transport robot;

[0080] If the actual number of transport robots is configured The harvesting robots will inevitably experience downtime and waiting, increasing the number of transport robots required, thus affecting the actual number of transport robots deployed. ; At that time, the theoretical downtime approaches zero.

[0081] The dynamic adjustment mechanism for the number of transport robots: as described above Follow It updates in real time based on dynamic changes, and This refers to the round-trip time for all completed trips. The statistical average value directly reflects the average distance from the currently completed fruit trees to the unloading point. As harvesting operations gradually advance from the depths of the orchard towards the unloading point, the average distance to subsequent fruit trees gradually shortens. It then decreased. The system can thus dynamically adjust the number of transport robots: in the early stages of operation (when the fruit trees are far from the unloading point and the round trip takes a long time), more transport robots are automatically configured to ensure continuous harvesting; as the operation progresses to the vicinity of the unloading point (where the round trip time is significantly reduced), If the level is reduced to a lower level, the investment in transport robots can be appropriately reduced, and surplus vehicles can be relocated or placed on standby, thereby optimizing equipment utilization and reducing energy consumption during idle runs. Conversely, if the operation path extends to a more distant area, leading to... rise, The system will also automatically adjust the number of transport robots deployed, ensuring that the downtime of the harvesting robots remains close to zero throughout the entire operation cycle.

[0082] (2) Four types of dynamic connection scheduling judgment situations:

[0083] Scenario 1: The current fruit tree has been harvested, but the fruit box has not reached the full load threshold. The picking robot moves to the next target fruit tree, and the current transport robot continues to follow and connect.

[0084] Scenario 2: The current fruit tree harvest is completed and the fruit box is just full (that is, the current transport robot's cumulative load has reached the preset load threshold). The current transport robot drives to the unloading point. The main control module sends a scheduling instruction to the transport robot in the idle state through the communication network, driving it to the next target fruit tree to wait for pick-up.

[0085] Scenario 3: The fruit trees have not been harvested yet, the fruit boxes are full in advance, the current transport robot is heading to the unloading point, and the main control module dispatches an idle transport robot to the current fruit tree location to take over the work through the communication network;

[0086] Scenario 4: All transport robots are fully loaded and heading to the unloading point. The harvesting robots pause their harvesting operations until any transport robot completes unloading and resets, at which point the harvesting operation resumes.

[0087] The order of priority determination for the three levels of dynamic right-of-way in step 3 is as follows:

[0088] Level 1 - Load Status Priority: Transport robots that are fully loaded and heading to the unloading point have higher priority than empty, standby, or connecting transport robots.

[0089] Level 2 - Target Distance Priority: Under the same load conditions, the transport robot that is closer to its own target point has priority to pass;

[0090] Level 3 - Task Type Priority: Under the same load and target distance, the transport robot that goes to the fruit tree location to take over the work has higher priority than the vehicle that returns to the waiting area.

[0091] Once the priority is determined, the main control module of the transport robot with lower priority automatically retrieves the nearest unoccupied preset avoidance point from the electronic map as the avoidance target, and drives the walking mechanism to enter the avoidance point. The preset avoidance points include road widening and intersections, and their occupancy status is synchronized in real time among multiple robots through the communication network to avoid multiple vehicles competing for the same avoidance point.

[0092] This embodiment uses the harvesting operation of mango orchards in the hilly areas of southern China as an application scenario to further explain the system and implementation method of the present invention.

[0093] 1. System Hardware Group

[0094] The system includes one harvesting robot and two transport robots of the same specifications. All robots are equipped with WiFi / 5G dual-mode wireless communication modules to achieve millisecond-level status synchronization.

[0095] The hardware configuration of the transport robot is as follows: a four-wheel independent drive walking mechanism; a fruit storage unit with a built-in high-precision weight sensor at the bottom; an integrated fruit box locking device that integrates electromagnetic locks and self-locking slots; a GNSS / INS positioning and attitude detection module and a radar obstacle avoidance module; a main control module and a wireless communication module.

[0096] The unloading point is equipped with a fruit box replacement station, and the waiting area serves as a standby point for empty transport vehicles. The integrated fruit box locking device automatically locks and triggers a micro switch after the empty box is replaced, sending a signal to the main control module that the unloading is complete, eliminating the need for manual remote issuance of return instructions.

[0097] 2. Example of Quantitative Calculation of Minimum Quantity of Transport Vehicles

[0098] Setting operating parameters: Harvesting efficiency of the harvesting robot The full load capacity of the transport vehicle ;

[0099] Operating Condition 1 (Short-distance transportation): Average round-trip time ,at this time:

[0100]

[0101] The need can be met by configuring only one transport robot;

[0102] Working Condition 2 (Large-area orchard in hilly terrain): Average round-trip time ,at this time:

[0103]

[0104] A minimum of two transport robots are required. If only one is configured, the harvesting robot will be continuously idle and waiting.

[0105] 3. Dynamic connection scheduling execution process

[0106] Real-time data collection of fruit box load and individual fruit tree harvesting progress, with adaptive scheduling across four scenario categories:

[0107] Scenario 1: After a single mango tree has been harvested but the fruit box is not full, the transport robot follows the picking robot to the next fruit tree to continue receiving fruit.

[0108] Scenario 2: Once a single fruit tree has been harvested and the fruit box is full, the current transport robot heads to the unloading point, while an idle transport robot is dispatched to the next fruit tree to wait.

[0109] Scenario 3: If a single fruit tree is not fully harvested and the fruit box is already full (a high-frequency scenario for mango harvesting), the fully loaded transport robot returns to unload, and the idle transport robot goes directly to the current fruit tree location to take over, with the picking robot only pausing for a few seconds.

[0110] Scenario 4: Both transport robots are fully loaded, and the harvesting robot suspends harvesting until either vehicle is unloaded and reset, after which operations resume.

[0111] 4. Multi-machine narrow-channel interference avoidance

[0112] The mango orchard road is a narrow, single-lane passage. When two vehicles are traveling in opposite directions, the main control module simultaneously reads the coordinates and speeds of the two transport robots and calculates the relative distance. Relative velocity And solve for the predicted collision time. .

[0113] Set a safety threshold: minimum safe distance Minimum safe time ;when or The avoidance logic is triggered at that time.

[0114] Traffic access is determined according to a three-level road right-of-way priority system. Low-priority vehicles are allowed to park and wait at the nearest unoccupied side road or widened yield point on the map, while high-priority vehicles have priority to pass. The occupancy status of yield points is synchronized in real time across the entire system to avoid competition for these points. When the lidar collision avoidance module detects non-cooperative obstacles such as tree branches or pedestrians, it directly executes the highest priority emergency braking.

[0115] 5. Slope and attitude adaptive safety control

[0116] GNSS / INS outputs vehicle attitude angles at a frequency of 100 Hz. In this embodiment, safety thresholds are set as follows: pitch angle (gradient) safety threshold is 12°, and roll angle (tilt) safety threshold is 8°.

[0117] If the roll angle of the transport robot reaches 9° (exceeding the limit), the transport robot immediately switches to low-speed crawl mode and simultaneously starts online replanning using the A* algorithm. The path cost function is weighted by the slope, and an alternative route is generated to bypass the gentler road section (the specific implementation method is the existing technology). If the same road section exceeds the attitude limit 3 times in a row, it is marked as a permanently dangerous road section in the global electronic map, and all subsequent transport robot path planning will automatically avoid it.

[0118] Meanwhile, the driving torque and regenerative braking force of the four-wheel independent drive walking mechanism are dynamically adjusted according to the uphill or downhill posture: the driving force is increased to prevent the vehicle from slipping when going uphill, and the braking force is increased to limit the vehicle speed when going downhill, ensuring the stability of the fully loaded fruit boxes.

[0119] 6. Multi-machine collaborative operation timing theory

[0120] T0~T1 stage: The harvesting robot and transport robot A are paired and follow each other to start the harvest;

[0121] Time T1: Transport robot A reaches the full load threshold, triggering scenario three scheduling. A drives to the unloading point, and transport robot B departs from the waiting area to the current fruit tree.

[0122] T1~T2 phase: The picking robot pauses briefly (about a few seconds), and resumes harvesting after the transport robot B arrives; the transport robot A completes the fruit box replacement at the unloading point, receives the reset signal and returns to the waiting area;

[0123] T2~T3 stage: Transport robot B has finished harvesting and is fully loaded, triggering scenario two scheduling. B goes to the unloading point, and transport robot A is scheduled to wait at the next fruit tree location.

[0124] Through repeated cycles, the downtime of the harvesting robot is reduced to just a few seconds; compared to the traditional single transport vehicle mode, the waiting time is reduced by about 67%, and the empty driving mileage is reduced by 40% to 60%.

[0125] 7. Security protection against communication and positioning failures

[0126] If communication is interrupted for more than 3 seconds or the positioning signal is lost, the transport robot will immediately cut off the drive power, slow down and park, and activate the audible and visual alarm. After the signal is restored, the vehicle will return to the nearest road network node or waiting area in low-speed safety mode and rejoin the task queue.

[0127] This embodiment fully verifies the entire process of transport vehicle quantitative configuration, dynamic shuttle scheduling, multi-machine interference avoidance, slope attitude protection, and fault safety protection, adapting to the complex operating environment of hilly mango orchards. Field test results show that the dual-transport vehicle configuration significantly improves harvesting efficiency and greatly reduces equipment idle time and the risk of vehicle collisions and rollovers.

[0128] This embodiment is only a preferred embodiment of the present invention. Equivalent substitutions and parameter adaptations made by those skilled in the art for other orchard scenarios such as citrus and apples without departing from the core concept of the present invention all fall within the protection scope of the present invention.

Claims

1. A dynamic, non-scheduling, multi-transport robot system for single-harvest mining, characterized in that, The system includes one harvesting robot and at least two transport robots with equivalent configurations. The robots communicate with each other via a bidirectional data transmission network. Each transport robot is equipped with a walking mechanism, a fruit storage unit, a main control module, and a weight sensor, a positioning and attitude detection module, a radar obstacle avoidance module, and a wireless communication module electrically connected to the main control module. The weight sensor is located at the bottom of the fruit storage unit and is used to collect the cumulative weight data of the fruit inside the storage box in real time. The positioning and attitude detection module integrates a GNSS / INS combined navigation unit and outputs the robot's global coordinates, speed, heading angle, roll angle, and pitch angle in real time. The radar obstacle avoidance module uses lidar to identify obstacles. The wireless communication module is used for synchronous sharing of position, speed, load status, task type, obstacle avoidance point occupancy status, and control commands among multiple robots.

2. The dynamic, non-scheduling, multi-transport robot system for single-harvest mining according to claim 1, characterized in that, The fruit storage unit integrates a single-unit fruit box locking device. This device includes an electromagnetic latch, a self-locking mechanical bracket, and a micro switch. The micro switch is located inside the self-locking mechanical bracket. The electromagnetic latch is used to release a full fruit box or lock an empty box under the control of the main control module. The self-locking mechanical bracket is used to automatically limit and securely lock the fruit box when an empty box is placed inside. After the transport robot finishes unloading and replaces the empty box, the self-locking mechanical bracket automatically locks the fruit box and presses the micro switch, sending an unloading completion electrical signal to the main control module.

3. The dynamic, non-scheduling interference-free single-collection multi-transport robot system according to claim 1, characterized in that, The main control module has built-in dynamic connection scheduling, multi-machine dynamic interference avoidance, and slope attitude safety protection; The dynamic shuttle scheduling is as follows: based on whether the load data of the weight sensor has reached the preset load threshold, and combined with the harvesting progress of the picking robot, a takeover instruction is dynamically generated, and an idle transport robot is dispatched to the target coordinate for seamless takeover. The multi-machine dynamic interference avoidance is as follows: real-time acquisition of the position, driving direction and speed information of surrounding transport robots, calculation and prediction of collision time, and then determination of whether there is an interference risk; when it is determined that there is an interference risk, the passage order of multiple transport robots is determined according to the preset dynamic right-of-way priority rules, and the transport robot with lower priority is driven to drive into the preset avoidance point to park and wait. The slope attitude safety protection is as follows: continuously collect the roll angle and pitch angle of the transport robot. If the detected value exceeds the preset safety threshold, the main control module immediately controls the transport robot to perform speed-limited crawling and triggers online path replanning to generate an alternative driving path.

4. The dynamic, non-scheduling interference-free single-collection multi-transport robot system according to claim 3, characterized in that, The dynamic shuttle scheduling includes quantitative calculation of the number of transport vehicles and four types of operation judgment scenarios; The specific calculation of the number of transport vehicles is as follows: Before operation, the round-trip time of the transport robots is estimated based on the orchard area, fruit tree distribution, and unloading points, and the initial configuration number is set; during operation, when the transport robots are fully loaded and scheduling is triggered, the shortest travel distance is calculated using the current fully loaded point as the starting point and the unloading point as the ending point. Combined with the average travel speed of the transport robot Time spent on a single unloading Calculate the time for a single round trip. The statistics show the round-trip time for all completed trips by the transport robots. This allows for the determination of the average round-trip time. Find the minimum number of transport vehicles required. ,in, This represents the average harvesting efficiency of the harvesting robot. This represents the full load capacity of the transport robots; if the actual number of transport robots configured... Increase the number of transport robots to achieve the desired actual number of transport robots deployed. ; The four types of task determination scenarios are as follows: (1) The current fruit tree has been harvested and the fruit box has not reached the full load threshold. The picking robot moves to the next target fruit tree and the current transport robot continues to follow and connect. (2) The fruit trees have been harvested and the fruit boxes are full. The current transport robot is heading to the unloading point. The main control module sends a scheduling instruction to the transport robot that is in an idle state through the communication network, driving it to the next target fruit tree to wait for pick-up. (3) If the fruit trees have not been harvested yet and the fruit boxes are full in advance, the current transport robot will drive to the unloading point. The main control module will use the communication network to dispatch an idle transport robot to the current fruit tree location to take over the operation. (4) All transport robots are fully loaded and head to the unloading point. The picking robots pause their harvesting actions until any transport robot completes unloading and resets, after which the harvesting operation resumes.

5. The dynamic, non-scheduling, multi-transport robot system for single-harvest mining according to claim 3, characterized in that, During the multi-machine dynamic interference avoidance process, the relative distance between two transport robots traveling in opposite directions is calculated using positioning data. Relative speed is calculated by differentiating continuous distance measurements. Solve for the predicted collision time ;like or The assessment indicated a risk of interference, among which, The minimum safe distance threshold is set. This is the minimum safe time threshold.

6. The dynamic, non-scheduling interference-free single-collection multi-transport robot system according to claim 3, characterized in that, The dynamic right-of-way priority rules are as follows: The first level is based on load status priority, where a fully loaded transport robot heading to the unloading point has a higher priority than a fully loaded transport robot; the second level is based on distance from the target, where, when the load status is the same, the transport robot that is closer to its target point has a higher priority. The third level prioritizes task type. When load and distance conditions are the same, the transport robot that goes to the fruit tree docking point to take over the work has a higher priority than the transport robot that returns to the docking waiting area to wait.

7. The dynamic, non-scheduling interference-free single-collection multi-transport robot system according to claim 6, characterized in that, Once the priority is determined, the main control module of the transport robot with the lower priority automatically retrieves the nearest unoccupied preset avoidance point from the electronic map as the avoidance target, and drives the walking mechanism to enter the avoidance point; the preset avoidance points include road widening sections and intersections, and their occupancy status is synchronized in real time among multiple robots through the communication network.

8. The dynamic, non-scheduling interference-free single-collection multi-transport robot system according to claim 1, characterized in that, The system is also equipped with fail-safe protection logic: when the communication interruption time of any transport robot exceeds a preset threshold or the positioning posture detection module signal is lost, the main control module cuts off the driving power of the walking mechanism, the transport robot slows down and parks in place and activates the audible and visual alarm; after the communication or positioning signal is restored, the transport robot returns to the nearest road network node or the docking waiting area in a low-speed safety mode.

9. A method for implementing a dynamic, non-scheduling, multi-transport robot system based on any one of claims 1-8, characterized in that, include: Step 1: The picking robot and multiple transport robots complete power-on self-tests, confirm that each transport robot is equipped with an empty fruit box and parked in the waiting area, and establish a multi-machine real-time location and status sharing communication network. Step 2: The picking robot performs the picking operation, and a designated transport robot follows and connects with it; the connection scheduling is dynamically adjusted according to the four types of operation judgment conditions, and the fully loaded transport robot drives to the unloading point, and the idle transport robot seamlessly takes over; Step 3: When multiple transport robots are traveling in opposite directions on the same road segment, the main control module calculates the relative distance and predicted collision time of the oncoming vehicles in real time during the robot's travel. If an interference risk is predicted, the low-priority vehicle is automatically driven to the nearest preset avoidance point and parked according to the dynamic right-of-way priority rules, while the high-priority vehicle has priority to pass. When multiple transport robots are traveling in the same direction on the same road segment, the main control module maintains a safe following distance and does not trigger interference avoidance. When a rear transport robot approaches a front transport robot, the front transport robot actively yields at the preset avoidance point. Step 4: Collect the vehicle posture of the transport robot in real time during the driving process. When the preset safety threshold is exceeded, the vehicle will crawl at a limited speed and replan to generate an alternative driving path. Step 5: After the fully loaded transport robot arrives at the unloading point, it uses the integrated fruit box locking device to replace the fruit box and automatically triggers the unloading completion electrical signal. The main control module then drives the transport robot to re-enter the transfer task queue. Step six: Repeat steps two through five until the harvesting task is completed.

10. The implementation method according to claim 9, characterized in that, If a vehicle's posture exceeds the limit multiple times on the same road segment, it will be marked as a permanently dangerous road segment, and the route planning will automatically avoid this road segment.

Images