LOW POWER CONSUMPTION BASED ON ADAPTIVE SENSOR FUSION. AUTOMATIC WASTE SORTING SYSTEM

Abstract

This invention relates to a low-power automated waste sorting system based on adaptive sensor fusion that enables the automatic recognition, classification, and redirection of different types of waste to appropriate compartments. The invention utilizes a hybrid decision-making mechanism based on image processing infrastructure, a low-power artificial intelligence processing architecture, and the combined evaluation of multiple sensor data. The system performs preliminary classification by analyzing image data of the waste, and depending on the achieved confidence level, it activates relevant sensors in stages to perform additional verification processes. This prevents unnecessary sensor usage, reduces processing load, and lowers energy consumption. The invention also enables the automatic transfer of waste to the appropriate compartments through mechanical guidance structures that operate according to the type of waste. The system is configured to operate in real-time on embedded systems with low processing power thanks to low-resolution image processing, data evaluation based on sensor reliability coefficients, confidence score calculation, and adaptive sensor activation. This results in lower energy consumption, lower hardware costs, and higher operational efficiency compared to existing systems requiring high processing power. The invention can be applied in various fields such as household use, offices, public spaces, recycling collection centers, and smart city applications, contributing to the automation of recycling processes and improving the accuracy of waste sorting.

Description

1 TARIFF 5 LOW POWER CONSUMPTION BASED ON ADAPTIVE SENSOR FUSION. AUTOMATIC WASTE SORTING SYSTEM Technological Field: This invention encompasses waste management, recycling technologies, embedded artificial intelligence systems, and sensors. Fusion-based decision support systems and electromechanical automation technologies 10 It relates to the following areas: More specifically, embedded systems with low processing capacity. with artificial intelligence algorithms optimized to work on hardware Waste is determined based on the combined evaluation of data obtained from multiple sensors. by determining the type and performing mechanical guidance according to the determined waste type. It relates to a low-power waste sorting system that performs automatic sorting. 15 The invention also relates to image processing technologies, weight and metal detection sensors, and adaptive imaging. sensor activation, real-time data processing architectures, energy-efficient decision-making hybrid systems where algorithms and motor-controlled steering mechanisms work together It includes a control infrastructure. The system is particularly suitable for; domestic use, smart cities. applications, office-type recycling systems and semi-industrial waste collection 20 solutions for the need for low-cost, compact and energy-efficient automated sorting It was developed for this purpose. State of the Art: Today, increasing population, changing consumption habits and recycling With the increasing importance of these requirements, automatic waste classification 25 Systems for their separation and decomposition are being widely developed, especially for plastics. Image processing-based system for separating metal, glass, and organic waste. artificial intelligence systems, sensor-assisted sensing structures, and mechanical guidance. Various technological solutions involving different mechanisms are being proposed. However, A significant portion of existing systems require high processing capacity, 30 It operates inefficiently in terms of energy consumption or only supports certain sensor types. due to their dependence on a sufficient level of accuracy in real-world usage conditions It is not accessible. When current technical solutions are examined, waste recognition, especially image processing-based solutions, is particularly noteworthy. It is observed that these systems are widely used. 35 systems have been developed within this scope. In these systems, camera data is analyzed through artificial intelligence algorithms, and waste is processed. The type is determined solely based on visual characteristics. However, these systems... 2 Due to the need for high-resolution image processing, powerful processors and high memory are required. It requires high capacity and intensive energy consumption. This is especially true for domestic use. This presents a challenge in implementation for low-cost embedded systems intended for general use. Furthermore, in systems that operate solely based on image data; dirty, crushed, deformed damaged, partially visible or light-affected waste should be properly treated. Classification becomes difficult. 10 In some known systems, in addition to the image processing structure, there is also a weight sensor and a metal sensor. or additional sensing components such as proximity sensors are used. However, these systems In the vast majority of cases, data obtained from sensors are independently analyzed. is being evaluated, and an optimized hybrid that analyzes sensor data together is being developed. There is no decision-making mechanism. Continuously operating the sensors is unnecessary. 15 This leads to increased energy consumption and processing load, especially at low power levels. In embedded systems targeting consumer consumption, this is a significant technical disadvantage. It constitutes. When current solutions are examined in terms of mechanical separation systems, mostly fixed guide flaps, simple servo mechanisms or single-axis 20 It is observed that guidance structures are used. However, in these mechanical systems... Adaptive steering based on waste type, jam detection, low power consumption movement. an optimized system that works integrated with control or real-time decision-making mechanisms The routing infrastructure is inadequate, especially for domestic use. In the compact systems developed, mechanical structures have limited mobility. This can lead to problems such as misdirection and mechanical jamming. It can open. In addition, a significant portion of existing systems have high processing capacity. need for centralized computer systems or cloud-based data processing infrastructures This situation increases both the system cost and the real-time 30 This can negatively affect the work. Also, constantly running high processing loads Artificial intelligence models increase energy consumption, making systems more portable. This limits its scalability and long-term performance. When the existing solutions within the known state of the art are considered together; It can run on embedded systems with low processing power, and its energy consumption is 35. minimizing, activating relevant sensors only when necessary, multiple evaluating sensor data together with a hybrid sensor fusion approach and the resulting 3 5 that automatically performs the mechanical sorting process according to the classification result. A compact, low-cost, and highly accurate waste sorting system is needed. It is heard. This invention was developed to overcome the technical problems mentioned above. It features a low-power AI infrastructure, adaptive sensor activation, and hybrid sensors. Thanks to its fusion-based decision-making mechanism and automated mechanical steering system, 10 It deviates from existing technical solutions. The purpose of the invention: The primary purpose of this invention is to address the problems that arise in domestic, office, and semi-industrial applications. automatic recognition of waste and its separation by directing it to the appropriate compartments a low power consumption, compact and high accuracy waste sorting system 15 It is about developing. Another aim of the invention is to enable processing of embedded hardware with limited processing capacity. high processing power using artificial intelligence algorithms optimized to work energy consumption, processing load and hardware compared to existing systems that require power The goal is to reduce costs. In this context, the system includes low-resolution image processing, 20 with optimized data analysis and decision-making mechanisms suitable for low resource consumption It is configured to work. The invention also addresses issues that arise in classification methods based solely on image data. To reduce misidentification problems, weight and metal parameters are collected along with camera data. and preferably a hybrid sensor 25 that evaluates the odor sensor data together. It aims to offer a decision-making mechanism based on fusion. Thus, dirty, Higher risk for deformed, partially visible, or light-affected waste. The aim is to classify them with accuracy. Another aim of the invention is to ensure that all sensors in the system are continuously active. 30 by preventing its operation and activating the relevant sensors only when necessary. Thanks to its adaptive sensor activation structure, it reduces energy consumption and improves process efficiency. The aim is to increase its efficiency. Thanks to this structure, the system, based on the preliminary assessment results, By progressively activating additional sensors based on the confidence score, unnecessary steps are eliminated. It minimizes the burden. The invention also includes motor-controlled mechanical steering that operates according to the specified waste type. 35 automatic sorting via the system without the need for user intervention. It aims to achieve this. In this context, the system includes a directional cover, a rotating 4 via drum mechanism or similar electromechanical steering structures 5 It ensures that the waste is transferred to the appropriate compartment. Another aim of the invention is to create low-cost standard electronic components and energy-efficient It can be scaled up using control infrastructure, requires low maintenance, and is suitable for mass production. The goal is to create a suitable automated waste sorting system. In addition, the system... its ability to connect to central data systems via wireless communication infrastructure, back 10 monitoring transformation data and ensuring integration with smart city applications. that is intended. Explanation of the Figures Figure 1: General perspective view of the automated waste sorting system that is the subject of the invention. Figure 2: View of the sensor placement on the system. 15 Figure 3: Block diagram of the hybrid sensor fusion and data processing architecture. appearance. Figure 4: Flowchart of the decision-making process based on adaptive sensor activation and confidence score. diagram view. Figure 5: View of the guidance system with mechanical guidance cover. 20 Figure 6: View of an alternative steering system with a rotary drum mechanism. Figure 7: Schematic view of the automated mechanical steering process. Figure 8: Schematic view of a low-power processing architecture. References: 1. Camera module 25 2. Image processing unit 3. Low power consumption artificial intelligence processing unit 4. Weight sensor 5. Metal sensor 6. Odor sensor 30 7. Waste inlet container 8. Mechanical steering flap 9. Rotary drum system 10. Motor driver 11. Microcontroller 35 12. Power supply 13. Wireless communication module 14. Data transfer system 5 15. Organic waste compartment 16. Plastic waste bin 17. Metal waste compartment 18. Sensor fusion module 19. Confidence score calculation unit: 10 20. User alert system 21. Energy management module 22. Sensor activation control unit 23. Mechanical transmission mechanism 24. Waste routing channel 15 25. Jamming detection sensor Description of the Invention: The invention concerns a low-power automated waste disposal system based on adaptive sensor fusion. The sorting system automatically recognizes, classifies, and separates different types of waste. An integrated recycling system developed to direct waste to appropriate compartments. 20 It is a system. The system consists of a camera module (1), an image processing unit (2), and a low power consumption system. artificial intelligence processing unit (3), weight sensor (4), metal sensor (5), odor sensor (6), waste inlet hopper (7), mechanical guiding cover (8), rotary drum system (9), motor driver (10), microcontroller (11), power supply (12), wireless communication module (13), data transfer system (14), organic waste compartment (15), plastic waste compartment (16), metal waste 25 compartment (17), sensor fusion module (18), confidence score calculation unit (19), user warning system (20), energy management module (21), sensor activation control unit (22), mechanical transmission mechanism (23), waste directing channel (24) and jamming detection It includes the sensor (25). Waste deposited into the system is transferred to the sensing area via the waste inlet hopper (7) 30 The camera module (1) located in the detection area obtains the image of the waste. by processing image data through the image processing unit (2) low power artificial transfers to the AI ​​processing unit (3). Low power AI processing unit (3) an optimized AI model working on it; low-resolution image analysis, shape determination, edge analysis and basic object classification processes 35 By doing this, it creates a preliminary classification result for the waste. Thus, high 6 Low processing power without the need for computer systems with high processing capacity. Real-time classification on embedded hardware with the capacity to do so. is being carried out. The system does not only work based on image data, but also via a weight sensor (4). Waste mass information, metal content information via metal sensor (5) and preferably used Organic content information is obtained via the odor sensor (6). 10 obtained from the sensors The data obtained are evaluated together within the sensor fusion module (18) and different sensor outputs, sensor reliability coefficients, and classification validation parameters They are evaluated in a weighted manner using these methods. Thus, dirty, crushed, deformed, affected by light conditions, or partially visible This ensures that waste is classified with a higher accuracy rate. 15 Confidence score calculation unit (19), classification obtained as a result of image processing By evaluating the data and the measurement results obtained from the sensors, a level of confidence is established. This level of confidence is determined by the probability of classification and sensor data. by considering the compatibility rate and data matching parameters between them This is determined. As a result of the preliminary classification, the confidence score is set at a certain threshold value of 20. If it remains below this level, the sensor activation control unit (22) activates the relevant sensors in stages. It activates them in this way. Thus, the system ensures that all sensors are constantly active. By preventing this, it reduces unnecessary energy consumption and processing load. The energy management module (21) is designed to control the power consumption of system components. It is structured. In this context, the energy management module (21) has 25 unused sensors. switching to passive mode, activating low-power operating modes and process It enables energy optimization based on power density. The source (12) is the energy of the electronic and electromechanical components within the system. It meets the need. After determining the type of waste, the microcontroller (11) drives the motor driver (10) 30 It operates the mechanical transmission mechanism (23) by controlling it. First application Mechanical guidance cover (8) in the form of waste corresponding to the relevant type of waste It is moved to align with the guiding channel (24) and the waste is appropriate The transfer to the compartment is ensured. In an alternative application method, a rotary drum is used. The system (9) is rotated so that it aligns with the relevant waste compartment. Thus, the waste; 35 into the organic waste compartment (15), plastic waste compartment (16) or metal waste compartment (17) It is automatically redirected. 7 The jamming detection sensor (25) located within the system detects during mechanical steering 5 It detects potential jamming or mechanical obstruction situations. Jamming If detected, the microcontroller (11) stops the motor driver (10) or reverses it. by creating a movement command in that direction, damage to mechanical system components It prevents. The user is given visual, auditory or warnings through the user warning system (20). Wireless notifications can be sent. 10 Thanks to the wireless communication module (13) and data transmission system (14), the system can be operated by The type of waste collected, fill level, usage intensity, and sensor data are sent to external devices. or they can be transferred to central data systems. This allows for the recycling processes to be streamlined. remote monitoring, data analysis, and integration with smart city infrastructures. It can be provided. 15 Sensor activation control unit (22), confidence obtained as a result of pre-classification. It is configured to activate the sensors gradually depending on the score. In this context, the system performs image analysis via the camera module (1) in the first stage. performing, the confidence score remaining below the predefined threshold value In this case, the weight sensor (4) and the metal sensor (5) are activated, organic content 20 If there is a need for further verification of the detection, the odor sensor (6) is activated. This prevents unnecessary sensor usage, thereby reducing processing load and energy consumption. Consumption is being reduced. Confidence score calculation unit (19), image classification probability, sensor data by evaluating the matching rate and sensor verification parameters together 25 It establishes the level of trust. The mechanical transmission mechanism (23) is controlled by the motor driver (10). will include gears, connecting rods, rotary shafts or linear motion elements that are moved. It can be configured in this way. The mechanical transfer mechanism (23) will reduce the redirection time and waste compaction. It can be configured with a controlled movement structure to prevent this. The system involves deactivating unused sensors and displaying low-resolution images. energy is applied to the process and only the necessary sensors are activated. It is structured in a way that will reduce its consumption. One of the key technical advantages of the invention is the hybrid sensor fusion, based on a confidence score of 35. High performance thanks to adaptive sensor activation and low power consumption data processing architecture. without the need for central computer systems with low processing capacity 8 High-accuracy real-time processing on embedded systems with processing capacity. The goal is to enable automated waste sorting. Industrial Application of the Invention The invention concerns a low-power automated waste disposal system based on adaptive sensor fusion. separation system; existing electronics manufacturing, plastic injection and electromechanical It can be produced in a way suitable for mass production using assembly technologies. System 10 Its body is manufactured using plastic injection, metal forming, or composite production methods. It can be manufactured, and the electronic components within the system are standard printed circuit boards. They can be integrated onto the cards. The invention utilizes a camera module, a weight sensor, and a metal sensor. microcontroller, low-power artificial intelligence processing unit, motor driver, and wireless 15 Components such as communication modules are standard electronics commonly found on the market. The components can be selected, and the system has a low-cost and widely producible structure. This is ensured. Mechanical guide hatch, rotary drum system and mechanical transmission mechanism; 20 using servo motors, stepper motors or similar electromechanical actuators It can be manufactured and scaled according to different capacity requirements. The system Additionally, because it can be designed in a modular structure, it can have extra compartments suitable for different waste categories. Alternatively, additional sensor units can be integrated into the system. The invention is suitable for homes, offices, restaurants, shopping malls, schools, and public spaces. in buildings, recycling collection points and smart city applications 25 It can be used, especially with its automated sorting structure that reduces user intervention. By increasing the efficiency of recycling processes, incorrect waste sorting It is possible to reduce and increase the rates of recyclable waste collection. is happening. The system also connects to central monitoring systems thanks to its wireless data transmission infrastructure. They can communicate and share data on occupancy rate, usage intensity and waste types. This feature allows for remote monitoring. Thanks to this feature, municipalities can track returns. effectively within transformation facilities and smart city management infrastructures It is available for use. Thanks to the invention's low power consumption design, energy requirements are reduced, and the embedded system 35 Because it can operate on a high-end platform, it eliminates the need for expensive computer infrastructure. 9 This is not being heard. This situation increases the commercial viability of the system and broadens it. 5 This makes it possible for it to be used by a wide range of users.

Claims

REQUIREMENTS 5 1. Low power consumption automatic waste sorting based on adaptive sensor fusion. Its feature is that it is a system; a camera module (1) that obtains an image of the waste, from the camera module (1) an image processing unit (2) that processes the received image data, image processing 10 Preliminary classification of waste type based on data received from unit (2) a low power artificial intelligence processing unit (3) that performs the physical waste at least one sensor that obtains data on its features (4, 5), data obtained from the sensors A sensor fusion module (18) that evaluates data together, classification confidence a unit of calculation of confidence score which forms the level (19), 15 depending on the confidence score a sensor activation control unit that activates the sensors in stages (22), a mechanical guiding cover that directs according to the type of waste. (8) or rotary drum system (9), the aforementioned mechanical guiding cover (8) or a motor driver (10) that controls the rotary drum system (9), system a microcontroller (11) that controls the components, motion 20 a mechanical transfer mechanism (23) that enables the transfer of wastes and the relevant It includes a waste redirection channel (24) that directs waste into the compartments and sensor activation control unit (22), obtained as a result of pre-classification The process activates the relevant sensors in stages depending on the confidence score. It should operate in a way that reduces its load and energy consumption. 25 2. Low power consumption automated waste management system based on adaptive sensor fusion according to Claim 1. It is a sorting system characterized by having at least one sensor that detects the weight information of the waste. a weight sensor (4) or a metal sensor that detects information about the metal content of the waste (5) is included.

3. Low power consumption automated waste disposal based on adaptive sensor fusion according to claim 1. It is a separation system and its feature is a low-power artificial intelligence processing unit. (3), low-resolution image processing, edge analysis, shape determination and basic It is configured to perform object classification operations.

4. Low power consumption automated waste management system based on adaptive sensor fusion according to Claim 1. The separation system has the following features: sensor fusion module (18), camera module 35 (1), data obtained from the weight sensor (4) and the metal sensor (5) using reliability coefficients and classification validation parameters It is a weighted evaluation. 11 5. Low power consumption automated waste disposal based on adaptive sensor fusion according to claim 1. The feature of the separation system is that the confidence score calculation unit (19), image classification probability, compatibility rate between sensor data, and data matching. It involves establishing a level of confidence by evaluating these parameters together.

6. Low power consumption automated waste management system based on adaptive sensor fusion according to Claim 1. The feature of the separation system is that the sensor activation control unit (22), the first 10 Activating the camera module (1) at this stage, the confidence score reaches the threshold value If it falls below, activate the weight sensor (4) and the metal sensor (5). This reduces unnecessary sensor usage and energy consumption.

7. Low power consumption automated waste disposal based on adaptive sensor fusion according to Claim 1. It is a separation system and its feature is that the system also detects organic content information 15 it includes an odor sensor (6) and the sensor activation control unit (22), additional If verification is needed, it activates the odor sensor (6).

8. Low power consumption automated waste disposal based on adaptive sensor fusion according to Claim 1. It is a separation system and its feature is that the system also controls energy consumption. It includes an energy management module (21) and the energy management module (21) 20 low power operating modes by deactivating unused sensors It is to activate it.

9. Low power consumption automated waste disposal based on adaptive sensor fusion according to Claim 1. The separation system is characterized by its mechanical transfer mechanism (23), mechanical 25 Controlled movement of the guiding cover (8) or the rotary drum system (9) gears, connecting rods, rotary shafts or linear motion elements that provide motion. It includes.

10. Low power consumption automated waste management based on adaptive sensor fusion according to Claim 1. It is a sorting system and its feature is; mechanical guiding cover (8), waste moving in such a way as to align with the guiding channel (24) the wastes to the relevant waste 30 Its purpose is to ensure that they are directed to their respective compartments.

11. Low power consumption automated waste management based on adaptive sensor fusion according to Claim 1. The feature of the separation system is that the rotary drum system (9) has the relevant waste compartment. rotating in such a way as to align the wastes into the organic waste compartment (15), plastic waste 35 12. Low power consumption automated waste management based on adaptive sensor fusion according to Claim 1. It is a separation system whose feature is that the system also handles mechanical jamming situations. it includes a jamming detection sensor (25) and jamming detection 12 If mechanical jamming is detected by the sensor (25), it sends 5 to the microcontroller (11) It is the process of sending data that will generate a control signal.

13. Low power consumption automated waste management based on adaptive sensor fusion according to Claim 1. It is a separation system and its feature is that the system also provides the user with visual, auditory or It includes a user alert system (20) that provides wireless notifications.

14. Low power consumption automated waste disposal based on adaptive sensor fusion according to claim 10 It is a data separation system; its feature is that the system also interacts with external devices via data. a wireless communication module (13) that enables communication and data transmission processes It includes a data transfer system (14) that performs and the type of waste, occupancy rate, usage intensity and sensor data transmitted to external devices or central data center The goal is to ensure that it is transferred to their systems. 15 15. Low power consumption automated waste disposal based on adaptive sensor fusion, according to Claim 1. It is a separation system; its feature is that the system also directs organic waste. It includes a waste compartment (15), a plastic waste compartment (16) and a metal waste compartment (17).

Images