WO2023086046A1

WO2023086046A1 - A feed pushing robot

Info

Publication number: WO2023086046A1
Application number: PCT/TR2022/050110
Authority: WO
Inventors: Ferdi ALAKUŞ
Original assignee: Itech Robotik Otomasyon Yenilenebilir Ener.Egt.Med.Taah.San.Ve Tic.Ltd.Sti.
Priority date: 2022-02-07
Filing date: 2022-02-09
Publication date: 2023-05-19

Abstract

The invention relates to a feed pushing robot that can be used to feed animals automatically without human power in the livestock sector.

Description

A FEED PUSHING ROBOT

TECHNICAL FIELD

DETAILED DESCRIPTION OF THE INVENTION

The invention relates to the improvements made on the system defined in our former patent application numbered TR2020/03200. The new features as a part of the improvement are described below.

The feed pushing robot, which is the subject of the invention, comprises a magnetic sensor, a camera, a sonar sensor, an encoder and a software that enables the technical effect of all these hardware.

The magnetic sensor is a board with 16 digital pins (outputs). Thanks to its 16 pins, it can read data from 16 sensors. The magnetic sensor informs the electronic card (industrial card) according to which input information (logic 1 ) comes from the sensors and tries to keep the motors at the ideal operating point according to the directions of these 16 sensors.

The magnetic sensor ensures that the robot is directed from the charging station where it is charging to the inside of the bam, which is the working area, and then detects the positions of the robot and calculates the rotation angles. Magnetic sensor is a card with a closed-loop control system that can control the movement without leaving any margin of error to the robot when crossing over an interruption on the road, when the robot comes to an end of the feed path, an end of the farm or a beginning of the farm; means in case of any interruption on its route, and in a case of the ongoing route after the interruption. For example, if there is a milking lane, the magnetic card ensures that the robot stops on the milking lane and crosses over without a danger. Thus, in farms with a milking lane, the robot can stop before reaching the milking lane while going straight and stops the helix and the pushing system while crossing over, then it turns 180 degrees to avoid damaging the feed pusher and proceeds via motion motors, and once the crossing over process is completed, all parts return to their normal working order. After crossing over the milking lane, the motion motors (movement) stop and turn 180 degrees again to reach the normal direction. Afterwards, movement and feed pusher motors are started and normal pushing procedure is continued. In this way, it is also prevented that the manure in the milking lane is mixed with the feed.

The robot stops with an obstacle detection switch in front of it when any big obstacle (human, large animal, etc.) comes up. In smaller obstacles, an image processing system is used, which comprises a method for processing the images taken by the camera (animal, child, etc.) so that the robot can stop by detecting the obstacles. Mentioned image processing is done via software. Image processing steps applied for object identification are given below.

- An image is taken through the camera.

- The taken image is resized to 256x256 pixels.

- A depth perception model is called and the resized image is sent as an input to the depth perception model.

- In the output image obtained from the depth perception model, the objects are colored according to their distance. The closest objects are colored white, the farthest objects are colored black and the other objects are colored between white & black based on their distance.

- In the resulting image from distance color schemes, values closer to black are turned completely black, with RGB color value being approximately 2 meters.

- The object detection model is called.

- The final image is sent as an input to the object detection model.

- By means of the object detection model, each pixel of the image is classified and compared with the previously trained object images.

- The coordinates and class of the object detected as a result of comparison and classification are returned as output.

- According to the coordinates obtained, the locations of these objects in the image are boxed according to the specified coordinates.

- The detected object class is displayed on each box.

- If a living object is detected, a "stop" command is sent to the machine.

- By making object identification, the system switches to a warning mode when any object enters within 2 meters. - As long as the detected object stops and the distance of 2 meters continues to decrease, it waits to detect a distance of 1 meter due to the approach of the robot to it. When a distance of 1 meter is reached, the robot slows down its speed. If the distance is half a meter, it stops completely.

The invention comprises two sonar sensors. These sonar sensors perform distance tracking without the need for GPRS data. The first sonar sensor controls the robot's primary motor to adjust its distance from the feeding fences (the place between the fence where animals stay and the feeding area). The speed adjustment (fast and slow) after the effect values can be adjusted by the user. The second sonar sensor monitors its distance from the first sonar sensor. As much as the difference, it controls the second motor so that the effective value of this sensor is controlled (by accelerating or decelerating it). In this way, it is ensured that the vehicle becomes parallel with the feeding fences. At the same time, it is possible to enter and exit from the narrow places around the columns in the barns. A suitable distance tracking can be done for each bam. One of the sensors is placed in front of the robot. Thus, for example, thanks to the sonar sensor on the helix side is in front of the machine, the columns in the bam are detected in advance and possible accidents are prevented. Since the effect values of the two sensors can be adjusted independently it can be ensured that the robot opens quickly and closes slowly according to the column or feeding fences.

The robot, which is the subject of the invention, comprises a multiple of current sensors. Thanks to this equipment, if the amount of feed in front of the robot is high, the speed of the motion motors can be slowed down and the amount of feed pushed per square meter by the helix system can be adjusted. Otherwise, if the amount of feed in front of the robot is not much, the robot can increase the motor speed with the current sensors. The density of the feed amount in front of the robot is determined by the camera.

The feed pushing robot, which is the subject of the invention, can make as many rounds as desired per day depending on the feed density, in certain distances (for example: between 40 and 100 cm) from the feeding fences. For example, when the first feeding is done, the amount of feed per square meter will be very high, so the robot can get closer starting from the outside at specified intervals in each round. After the first feeding, 100 cm in the first round, 80 cm in the second round, since the feed amount will decrease 60 cm in the third round and up to 40 cm close to the feeding fences in the last round before refeeding, in order to bring the feed to the ideal point for the cows to eat.

Our invention comprises an independent double-flighted helix. Double-flighted means that there are two contact surfaces in each helix loop. In this way, a better ventilation of the feed is ensured and frequent feed portions due to the smaller helix pitch is provided.

The distance traveled by the robot is determined by the encoder connected to the motors. It is calculated according to the motor’s revolution. The encoder also provides the calculation and recording of the latest position data. The distance traveled in each direction is recorded. Thus, the current position is determined relative to the starting point. In this way, it is ensured that the robot can operate from the same position again in cases where standard loops are interrupted due to factors such as stopping the robot by pressing its emergency stop button.

The software is the unit where all the calculations of the robot are made and the decision mechanism is run. It is ensured that the angle values of the robot are entered via the interface on the screen, the movements of the robot are performed according to these entered angle values and the controls are facilitated. With the control effect values (for example, angles) entered by the user, the sharpness and softness of the speed of the motors, the layout of the farm can be adjusted regarding if it is column-free or it has a route with or without rotation. Sensitive adjustments can be made according to the environment. With the values entered via the interface through the screen on the robot, the positioning of the charging station and passages between the farms can be determined. Again, via the interface, the operating hours of the robot and the distance to be pushed can be adjusted by the user, according to the feeding times (feeding once a day, twice a day, three times a day) and feed densities.

The data coming from the magnetic sensor is also processed through software and became usable. The magnetic positioning sensor comprises 16 holl-effect sensors. These sensor data can be received as RS 232 communication protocols and “logic” output. Positioning is done according to this received data. The logic data coming from magnetic sensors are processed on the electronic card. Each logic sensor is multiplied by 10 times its sequence number and divided by the incoming logic number to calculate the instantaneous position of the 25 mm wide (this may vary) magnetic tape on the sensor. The electronic card sends the necessary data for motion data (positioning) to the motor drivers and to the motors via the motor drivers according to the instant position in order to approach the position data previously set by the user via the screen. In this way, the robot follows the magnetic line by adjusting the speed of the two motors (by slowing one down and accelerating the other). Using the same sensor data, "magnetic line presence/absence" data is also obtained. According to this data, two motors can be stopped and started at the same time (For example, the robot stops at the end of the farm, the magnetic sensor stops or starts the motor when approaching the charging station, etc.).

During the rotation of the robot, the magnetic sensor is detected and followed, so that the rotation is successful. According to the values entered on the screen robot's angle values, times, motor controls, charging station location, maximum and minimum fence distance, working hours, fence distance during working hours, movement speed settings can be made in the electronic card. According to these adjusted data, the electronic card sends the necessary signals to the motor drivers. According to these sent signals, the motors are directed and the autonomous feed pushing robot is brought to the desired position.

With the camera, if the user desires, the images acquired during the movement of the robot can be watched live, thanks to the instant image transfer. These images can also be saved. Access to these images by the user is possible with the interface of the robot which provides remote/wireless access. There are wireless access protocols for this access on the robot.

Within the invention, there is a vitamin tank where the user can add vitamins to be mixed with the feed to be pushed. There is a helix connected to a helix drive motor in this tank. A portioning process can be performed at desired dosages with the helix in the vitamin tank. Substances such as vitamins and promix to be put from the vitamin tank by the user are distributed on the feed during pushing. According to the data coming from the current sensor, the feed density is determined by the software and the required amount of vitamin portion is adjusted automatically. The electronic card sends data to the motor in the vitamin tank via the motor driver for vitamin portioning. The electronic card uses the feed density data coming to the industrial card via the current sensor to increase or decrease the feed portioning by accelerating or slowing down the helix motor by changing the motor speed. Thus, the amount of vitamin poured from the vitamin tank to the square meter is changed.

Claims

CLAIMS ed pushing robot characterized in that comprising;

- A magnetic sensor further comprising 16 digital pins (outputs), reading data from 16 sensors, directing the robot from the charging station where it is charging to the inside of the bam, which is the working area and detecting robot’s positions and sending the necessary data for motion data (positioning) to the motor drivers and to the motors via the motor drivers according to the instant position in order to approach the position data previously set via the screen, calculating the rotation angles, further comprising a closed-loop control system that can control the movement without leaving any margin of error to the robot when crossing over an interruption on the road and stopping the helix and the pushing system while crossing over,

- Motion motors activated when the robot turns 180 degrees to avoid damaging the feed pusher and prevent the mixing of the manure in the milking lane with the feed while crossing over and once the crossing over process is completed, all the parts return to their normal working order,

- an obstacle detection switch in the front to detect if any big obstacle (human, large animal, etc.) comes up,

- two sonar sensors that the first one controls the robot's primary motor to adjust its distance from the feeding fences and the second one to monitor its distance from the first sonar sensor and adjusts the second motor by accelerating or decelerating it and to perform distance tracking without a need for GPRS data,

- a multiple of current sensors adjusting the amount of feed pushed per square meter through the helix system according to the amount of feed in front of the robot by slowing down the speed of the motion motors,

- an independent double-flighted helix,

- an encoder determining the distance travelled by the robot by the motor’s revolution, recording the travelled distance in each direction and determining the current position relative to the starting point,

6 - an electronic card further comprising a software processing the logic data coming from magnetic sensors, providing the entering of angle values of the robot and performing the movements of the robot according to these angle values, providing the entering of the information of the sharpness and softness of the speed of the motors, the layout of the farm if it is column-free or whether it has a route with or without rotation, the positioning of the charging station and passages between the farms, the operating hours of the robot and the distance to be pushed via interface through the screen according to the feeding times (feeding once a day, twice a day, three times a day) and feed and decision making by processing the images taken through the camera, detecting the feed density according to the data coming from the current sensor and adjusts the necessary vitamin portion for this by speeding up or slowing down the helix motor,

- And the wireless access protocols.

2. The feed pushing robot of Claim 1 characterized in that comprising a camera to provide instant image transferring to the user and providing an image to detect small obstacles (animal, child, etc.) by an image processing method via the software.

3. The image processing method of Claim 2 characterized in that comprising the steps below:

- Taking an image through the camera,

- Resizing the taken image to 256x256 pixels,

- Calling a depth perception model and sending the resized image as an input to the depth perception model,

- Coloring the objects in the output image obtained from the depth perception model according to their distance and coloring the closest objects white, coloring the farthest objects black and coloring the other objects between white & black based on their distance,

- Turning the values closer to black in the resulting image in which the distance color schemes are obtained, to completely black, with RGB color value being approximately 2 meters,

- Calling the object detection model,

7 - Sending the final image as an input to the object detection model,

- Classifying each pixel of the image by means of the object detection model, and comparing with the previously trained object images,

- Detecting the coordinates and class of the object and as a result of comparison and classification are returned as output,

- According to the coordinates obtained, the locations of these objects in the image are boxed according to the specified coordinates,

- Displaying the detected object class on each box,

- Sending a "stop" command to the robot if a living object is detected,

- By making object identification, the system switches to a warning mode when any object enters within 2 meters,

- As long as the detected object stops and the distance of 2 meters continues to decrease, waiting to detect a distance of 1 meter due to the approach of the robot to it,

- When a distance of 1 meter is reached, slowing the robot’s speed,

- If the distance is half a meter, stopping the robot completely. The sonar sensor of Claim 1 characterized in that the sonar sensor on the helix side is positioned in front of the machine and the effect values of the two sensors can be adjusted independently.

8