Object recognition system for autonomous vehicles

Abstract

An object detection system for autonomous vehicles, comprising the following: Several vehicle-mounted sensors, including at least one camera, at least one LiDAR sensor, at least one radar sensor, and at least one ultrasonic sensor, each sensor configured to capture data representing the environment surrounding the vehicle; a sensor fusion module functionally connected to the multiple sensors, the sensor fusion module being configured to receive and integrate data from the multiple sensors to produce a comprehensive representation of the vehicle's environment; a data processing unit operationally connected to the sensor fusion module, wherein the data processing unit is configured to preprocess the integrated data by filtering noise, highlighting relevant features, and normalizing sensor outputs to a common frame of reference; a feature extraction module operationally connected to the data processing unit, wherein the feature extraction module is configured to identify and extract characteristic features from the preprocessed data that correspond to objects within the vehicle environment; an object detection unit operationally connected to the feature extraction module, wherein the object detection unit comprises a deep neural network model configured to classify the extracted features in real time into predefined object categories such as vehicles, pedestrians, cyclists, traffic signs, and obstacles; and a decision controller that is operationally connected to the object recognition unit, wherein the decision controller is configured to analyze the classified objects, determine potential collision risks and generate appropriate control commands to adjust the navigation path, speed and other driving parameters of the vehicle to ensure safe and efficient autonomous operation.

Description

Technical area

[0001] The present invention relates to autonomous vehicle technology, particularly to a system for detecting objects in the vehicle's environment to enable safe and efficient autonomous navigation. The invention utilizes a combination of sensors, machine learning algorithms, and advanced data processing techniques to accurately identify and classify objects in real time. Background of the invention

[0002] Autonomous vehicles rely heavily on their ability to accurately perceive their surroundings and react accordingly to ensure safe navigation. Conventional object detection systems in autonomous vehicles use various types of sensors, such as LiDAR, cameras, radar, and ultrasonic sensors, to collect data from the environment. However, these systems often face challenges in complex urban environments where many objects of different sizes and types are present, such as pedestrians, cyclists, vehicles, traffic signs, and obstacles. Existing systems may suffer from limitations in sensor fusion, data processing speed, and accuracy, which can lead to incorrect or delayed responses. Therefore, there is a need for an improved object detection system that increases the safety and reliability of autonomous vehicles by overcoming these limitations.

[0003] In recent years, the development of autonomous vehicles has made great strides, driven by the potential benefits of increased safety, reduced traffic congestion, improved fuel efficiency, and enhanced accessibility. A key component of autonomous vehicle systems is their ability to detect and classify objects in their environment, which is essential for making precise driving decisions and ensuring the safety of passengers, pedestrians, and other road users. Current object detection technologies for autonomous vehicles typically rely on a combination of sensors such as cameras, LiDAR (Light Detection and Ranging), radar, and ultrasonic sensors, each of which has its own advantages and limitations.Despite advances in these technologies, existing object detection systems still face several significant challenges that affect their performance in real-world applications.

[0004] Camera-based systems are among the most commonly used object detection solutions in autonomous vehicles. These systems use digital cameras to capture high-resolution images of the vehicle's surroundings, which are then processed using computer vision techniques and machine learning algorithms such as convolutional neural networks (CNNs) to detect and classify objects. Cameras provide rich visual information, including color, texture, and contrast, which is valuable for detecting objects such as road signs, traffic lights, and pedestrians. However, camera-based systems have several disadvantages. They are highly sensitive to lighting conditions, and their performance can degrade significantly in low light, direct sunlight, shadows, or glare. In addition, cameras have limited depth perception, making it difficult to accurately estimate the distance to objects and judge their relative speeds.Cameras also have problems in adverse weather conditions such as rain, fog or snow, as visibility is limited and objects are difficult to detect.

[0005] LiDAR technology, which uses laser beams to measure distances and create detailed 3D maps of the surroundings, is another widely used object detection solution in autonomous vehicles. LiDAR sensors emit laser pulses and measure the time it takes for the pulses to reflect off objects and return to the sensor, creating a precise, high-resolution point cloud representation of the surroundings. LiDAR is particularly effective at detecting the shape and size of objects and is less affected by lighting conditions, making it ideal for nighttime driving. However, LiDAR systems have several limitations. They are relatively expensive, making them a costly component in the overall autonomous vehicle system. Furthermore, LiDAR performance can be degraded in adverse weather conditions.For example, heavy rain, fog, or snow can scatter the laser beams, causing noise in the data and reducing the accuracy of object detection. LiDAR systems also generate vast amounts of data, which requires significant computing resources to process in real time, which can be a limitation for vehicles with limited processing power. Furthermore, LiDAR sensors have a limited range of typically a few hundred meters, which may be insufficient for high-speed driving scenarios where longer detection ranges are required.

[0006] Radar-based systems are also commonly used in autonomous vehicles for object detection. Radar sensors emit radio waves and measure the time it takes for the waves to bounce back after hitting an object, providing information about the object's distance and relative speed. Radar is particularly effective at detecting objects in various weather conditions such as fog, rain, or snow, where camera and LiDAR systems may struggle. Radar is also less affected by lighting conditions and can detect objects at greater distances than LiDAR, making it suitable for high-speed driving. However, radar systems have their limitations. While radar can effectively detect the presence and speed of objects, it provides relatively low-resolution data compared to cameras and LiDAR, making it difficult to accurately determine the shape and type of objects.This limitation can lead to false positives or negative results, where harmless objects are mistakenly classified as hazards or actual hazards are missed. Radar systems can also be susceptible to interference from other radar systems, resulting in inaccuracies in object detection.

[0007] Ultrasonic sensors, which use sound waves for object detection, are often used for close-range object detection, for example, in parking assistance systems. These sensors are relatively inexpensive and can effectively detect objects at close range. However, due to their limited range, typically only a few meters, they are not suitable for object detection at the distances required for most autonomous driving scenarios. Ultrasonic sensors are also susceptible to ambient noise, such as noise from nearby vehicles or construction sites, which can affect their performance.

[0008] To overcome the limitations of individual sensors, many autonomous vehicle systems employ a multi-sensor fusion approach, which combines data from multiple sensors to gain a more comprehensive understanding of the vehicle's surroundings. Sensor fusion integrates data from different sensor types, such as cameras, LiDAR, radar, and ultrasonic sensors, to leverage the strengths of each while addressing their weaknesses. For example, sensor fusion can combine the high-resolution visual data from cameras with the accurate depth information from LiDAR and the robust detection capabilities of radar in adverse weather conditions. This approach aims to improve the overall reliability and accuracy of object detection.

[0009] However, multi-sensor fusion systems face their own unique challenges. One of the biggest is data synchronization, as each sensor operates at different frequencies and generates data with different latencies. For accurate object detection, it is critical to correctly align the data from multiple sensors in time and space. However, this process can be complex and computationally intensive. Furthermore, integrating data from sensors with different resolutions, formats, and noise characteristics requires sophisticated algorithms to effectively filter and fuse the data. Another challenge is the computational complexity of multi-sensor fusion systems, as they generate large amounts of data that must be processed in real time to enable timely decision-making.This requirement can place a strain on the vehicle's computing resources, especially in scenarios where computing power is limited or energy efficiency is a concern.

[0010] Machine learning and deep learning algorithms are increasingly being used for object detection in autonomous vehicles to improve the accuracy and robustness of these systems. Deep neural networks such as convolutional neural networks (CNNs) and recurrent neural networks (RNNs) have shown promise for improving object detection by learning complex patterns and features from large datasets. These algorithms can be trained to detect various objects such as pedestrians, vehicles, cyclists, and traffic signs with high accuracy. However, deep learning models also have disadvantages. They require extensive computational resources for training and inference, which can be challenging for real-time applications in autonomous vehicles. Furthermore, these models are often data-hungry and require large amounts of labeled training data to achieve high accuracy.In many cases, collecting and annotating such data is time-consuming and costly. Deep learning models can also be prone to overfitting, where the model performs well on training data but poorly on new, unfamiliar data, which can limit their generalization capabilities across different driving environments.

[0011] Furthermore, the interpretability of deep learning models is often limited, making it difficult to understand the model's decision-making process and ensure its reliability in safety-critical applications such as autonomous driving. Another challenge is the ability of these models to handle rare or unexpected events, such as unusual pedestrian behavior or objects that are not well represented in the training data. Autonomous vehicles must operate safely in all scenarios, even rare or anomalous ones, and current deep learning models may not always be equipped to handle such situations effectively.

[0012] Given the limitations and challenges of existing object detection technologies, there is a clear need for an improved system that enables reliable, accurate, and real-time object detection in a variety of driving environments and conditions. The present invention addresses these needs by combining advanced multi-sensor fusion techniques with state-of-the-art machine learning algorithms to create a robust and efficient object detection system for autonomous vehicles. By leveraging the strengths of different sensor types, optimizing data processing algorithms, and employing adaptive machine learning models, the invention aims to overcome the drawbacks of existing solutions and improve the overall performance and safety of autonomous vehicle systems. Summary of the invention

[0013] The present invention provides an object detection system for autonomous vehicles that integrates a multi-sensor fusion module with a deep learning-based object classification unit. The system is designed for real-time operation and processes data from multiple sensors simultaneously to accurately identify and classify objects in the vicinity of the vehicle. The invention utilizes advanced machine learning models trained on diverse datasets to improve detection and classification accuracy, particularly in challenging environments such as urban areas or adverse weather conditions.

[0014] The system consists of several key components: a sensor array with LiDAR, radar, cameras, and ultrasonic sensors, a multi-sensor fusion module, a deep learning-based object classification unit, a central processing unit (CPU) for data processing, a memory unit for storing data and algorithms, and a communication interface for transmitting data to the vehicle's control system.

[0015] The primary objective of the invention is to provide an advanced object detection system for autonomous vehicles that significantly improves the safety, accuracy, and reliability of navigation in diverse and challenging environments. This invention aims to overcome the limitations associated with conventional object detection systems by integrating multiple sensor modalities, including cameras, LiDAR, radar, and ultrasonic sensors, into a unified framework that leverages the strengths of each sensor type. By employing sensor fusion techniques, the system combines data from these various sensors to create a more comprehensive and accurate representation of the vehicle's environment, thus enhancing object detection and classification capabilities.

[0016] Another objective of the invention is to utilize advanced machine learning algorithms, such as deep learning models, to improve the system's ability to identify and classify various objects in real time. The use of sophisticated neural networks allows the system to learn from large data sets and improve its accuracy over time, especially in complex urban environments where objects can vary greatly in shape, size, and appearance. This adaptability ensures that the system can accurately detect and distinguish between pedestrians, cyclists, vehicles, traffic signs, and other potential obstacles, even in poor visibility or adverse weather conditions.

[0017] Furthermore, the invention is intended to improve the real-time processing capabilities of object detection systems in autonomous vehicles. By optimizing data processing algorithms and employing hardware acceleration techniques, such as GPUs or dedicated AI chips, the system ensures that the detection and classification processes are performed with minimal latency. This real-time processing capability is crucial for autonomous vehicles to make fast and accurate decisions in dynamic driving scenarios, thereby reducing the likelihood of accidents and improving overall safety.

[0018] Another objective of the invention is to provide a scalable and cost-effective solution for object detection in autonomous vehicles. By developing a modular and flexible system architecture, the invention enables easy integration into various types of autonomous vehicles, from passenger cars to commercial vehicles. This scalability ensures that the system can be adapted to different vehicle platforms and sensor configurations without significant changes, thereby reducing development costs and time to market. SHORT DESCRIPTION OF THE FIGURE

[0019] These and other features, aspects, and advantages of the present invention will become more readily understood when the following detailed description is read in conjunction with the accompanying drawings, in which like characters represent like parts throughout. Fig. Figure 1 shows a block diagram of a system for adaptive English reading comprehension training.

[0020] Furthermore, those skilled in the art will appreciate that elements in the drawings are shown for convenience and may not necessarily be drawn to scale. For example, the flowcharts illustrate the method by key steps to enhance understanding of aspects of the present disclosure. Furthermore, with respect to device construction, one or more components of the device may have been represented in the drawings by conventional symbols, and the drawing may show only the specific details relevant to understanding embodiments of the present disclosure in order not to clutter the drawings with details that would be readily apparent to those skilled in the art who would benefit from the description herein. Detailed description of the invention

[0021] To facilitate an understanding of the principles of the invention, reference will now be made to the embodiment illustrated in the drawings and described in specific language. It is to be understood, however, that no limitation on the scope of the invention is intended thereby, since such changes and further modifications of the illustrated system, and such further applications of the principles of the invention as illustrated therein, are contemplated as would normally occur to one skilled in the art to which the invention pertains.

[0022] It will be understood by one skilled in the art that the foregoing general description and the following detailed description are exemplary and explanatory of the invention and are not intended to be limiting thereof.

[0023] References in this specification to "one aspect," "another aspect," or similar expressions mean that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the present disclosure. Therefore, occurrences of the terms "in one embodiment," "in another embodiment," and similar expressions throughout this specification may or may not all refer to the same embodiment.

[0024] The terms "comprises," "comprising," or other variations thereof are intended to cover non-exclusive inclusion, such that a process or method comprising a list of steps not only includes those steps, but may also include other steps not expressly listed or inherent in such process or method. Likewise, one or more devices or subsystems or elements or structures or components preceded by "comprises...a" does not preclude, without further limitation, the existence of other devices or other subsystems or other elements or other structures or other components or additional devices or additional subsystems or additional elements or additional structures or additional components.

[0025] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. The system, methods, and examples provided herein are for illustrative purposes only and should not be considered limiting.

[0026] Embodiments of the present disclosure will be described in detail below with reference to the accompanying drawings.

[0027] The present invention creates a comprehensive traffic dataset to train and evaluate autonomous vehicle (AV) algorithms using objects specific to specific driving conditions to achieve safe and effective navigation. High-quality videos captured with dashboard cameras are carefully annotated using computer vision techniques. The annotations include detailed metadata such as vehicle types, road infrastructure features, and traffic signals. Furthermore, the dataset is validated using YOLO v5 to identify traffic-related objects.

[0028] With reference to Fig.10 illustrates a block diagram of an object detection system for autonomous vehicles, comprising: a plurality of vehicle-mounted sensors 102, including at least one camera, at least one LiDAR sensor, at least one radar sensor, and at least one ultrasonic sensor, each sensor configured to collect data representative of the vehicle's surroundings; a sensor fusion module 104 operatively connected to the plurality of sensors, the sensor fusion module configured to receive and integrate data from the plurality of sensors to generate a comprehensive representation of the vehicle's surroundings;a data processing unit 106 operatively connected to the sensor fusion module, wherein the data processing unit is configured to preprocess the integrated data by filtering noise, highlighting relevant features, and normalizing sensor outputs to a common reference frame; a feature extraction module 108 operatively connected to the data processing unit, wherein the feature extraction module is configured to identify and extract characteristic features corresponding to objects within the vehicle environment from the preprocessed data; an object detection unit 110 operatively connected to the feature extraction module, wherein the object detection unit comprises a deep neural network model configured to classify the extracted features in real time into predefined object categories such as vehicles, pedestrians, cyclists, traffic signs, and obstacles;and a decision controller 112 operatively connected to the object detection unit, the decision controller configured to analyze the classified objects, determine potential collision risks, and generate appropriate control commands to adjust the navigation path, speed, and other driving parameters of the vehicle to ensure safe and efficient autonomous operation.;

[0029] In one embodiment, the present invention further includes a calibration module 114 operatively connected to the plurality of sensors 102, wherein the calibration module is configured to perform real-time calibration of each sensor to correct for sensor drift, misalignment, and environmental changes, thereby ensuring accurate and consistent data acquisition from the plurality of sensors.

[0030] In one embodiment, the sensor fusion module 104 uses a probabilistic data fusion algorithm that incorporates Bayesian inference techniques to account for uncertainties and variations in the data received from the various sensors, thereby producing a more robust and reliable representation of the vehicle's environment.

[0031] In one embodiment, the present invention further includes a hardware-accelerated processing unit 116 comprising at least one of a graphics processing unit (GPU) or tensor processing unit (TPU) operatively connected to the data processing unit and the object detection unit, wherein the hardware-accelerated processing unit is configured to perform parallel processing of complex computational tasks related to noise filtering, feature extraction, and object classification, thereby reducing system latency and enabling real-time decision making.

[0032] In one embodiment, the present invention further includes an environmental condition detection unit 118 operatively connected to the data processing unit, wherein the environmental condition detection unit is configured to analyze the data from the plurality of sensors to detect adverse weather conditions such as rain, snow, fog, or low light, and to dynamically adjust the data processing parameters and object detection thresholds to maintain high detection accuracy under varying environmental conditions.

[0033] In one embodiment, the decision control controller 112 further includes a dynamic object tracking module configured to track the movement of classified objects over time using an algorithm based on a Kalman filter or a particle filter, predict future positions and trajectories of moving objects, and adjust the vehicle's control commands to avoid potential collisions based on the predicted trajectories.

[0034] In one embodiment, the present invention further includes a secure communication interface 120 operatively connected to the decision-making controller, the secure communication interface configured to send and receive data between the autonomous vehicle and a remote server or other nearby autonomous vehicles, thereby enabling cooperative sensing and shared situational awareness to improve overall safety and traffic management in complex driving environments.

[0035] In one embodiment, the present invention further includes a machine learning training module 122 operatively connected to the object detection unit, wherein the machine learning training module is configured to continuously update and retrain the deep neural network model using new data collected from the autonomous vehicle during operation, thereby improving the accuracy and adaptability of the model to new objects and driving scenarios.

[0036] The object detection system for autonomous vehicles comprises a comprehensive sensor array including LiDAR, radar, cameras, and ultrasonic sensors strategically positioned around the vehicle. The LiDAR sensors provide high-resolution 3D point cloud data of the environment and capture the spatial properties of objects. Radar sensors are used for object detection at greater distances and in various weather conditions such as fog or rain. Cameras capture high-resolution images and videos and enable the recognition of visual features such as colors, textures, and shapes. Ultrasonic sensors detect nearby objects and assist with low-speed maneuvers such as parking.

[0037] The data collected by these sensors is transmitted to the multi-sensor fusion module. This module uses advanced algorithms to combine data from different sensors, leveraging the strengths of each while balancing their limitations. For example, fusion of LiDAR and radar data improves the detection accuracy of objects at different distances, while integrating camera data adds visual context to detected objects. The fusion module uses a Kalman filter-based approach to synchronize data from different sensors and eliminate noise or inconsistencies.

[0038] Once the data is combined, it is passed on to the deep learning-based object classification unit. This unit consists of a neural network model trained on large-scale datasets containing various objects commonly found in urban and rural environments, such as pedestrians, cyclists, vehicles, animals, traffic signs, and obstacles. The model uses convolutional neural networks (CNNs) to extract features from the sensor.

[0039] The present invention relates to an object detection system for autonomous vehicles specifically designed to improve the safety and reliability of autonomous navigation through the use of an advanced multi-sensor fusion approach combined with sophisticated machine learning algorithms. The system comprises a plurality of vehicle-mounted sensors, including at least one camera, at least one LiDAR sensor, at least one radar sensor, and at least one ultrasonic sensor. These sensors are strategically placed around the vehicle to collect comprehensive data representing the vehicle's surroundings from multiple perspectives.Each sensor type offers different advantages; for example, cameras provide high-resolution images useful for detecting visual details such as colors and textures, LiDAR sensors create precise 3D point clouds for accurate depth perception, radar sensors are effective for detecting the distance and relative speed of objects in various weather conditions, and ultrasonic sensors provide close-range distance measurements, particularly useful for low-speed maneuvering and parking.

[0040] The core of the invention lies in the sensor fusion module, which is operatively connected to the multitude of sensors. The sensor fusion module is responsible for receiving raw data from all sensors and integrating this data to create a unified, comprehensive representation of the vehicle's environment. This module uses a probabilistic data fusion algorithm that employs Bayesian inference techniques to effectively combine the data from the various sensors, each of which may have varying degrees of reliability and accuracy. The probabilistic approach enables the system to manage uncertainties and discrepancies in the sensor data and provide a more robust and reliable environmental model than a single sensor could achieve alone.

[0041] After the sensor fusion process, the integrated data is transferred to the data processing unit. This unit plays a crucial role in preprocessing the data by performing operations such as noise filtering, data normalization, and enhancing relevant features. Noise filtering algorithms such as median filtering or Gaussian smoothing are used to remove irrelevant or interfering signals from the sensor data that could otherwise lead to false detections or inaccurate object classification. Data normalization ensures that outputs from different sensors, each with its own native resolution and format, are converted into a common reference frame, enabling accurate comparison and integration of the sensor data.The preprocessing step also includes feature enhancement techniques that enhance critical details in the data, such as edges and contours, which are essential for reliable object detection.

[0042] The present invention relates to the field of autonomous vehicle technology, particularly to object detection systems that enhance the ability of autonomous vehicles to perceive their surroundings and make informed navigation decisions. The invention incorporates advances in multi-sensor fusion, machine learning algorithms, and real-time data processing to improve the accuracy and reliability of object detection in diverse driving environments. This invention is particularly relevant to applications where autonomous vehicles must operate safely in complex and unpredictable environments such as urban areas, highways, and adverse weather conditions.

[0043] The drawings and the foregoing description provide examples of embodiments. Those skilled in the art will recognize that one or more of the described elements may well be combined into a single functional element. Alternatively, certain elements may be separated into multiple functional elements. Elements from one embodiment may be added to another embodiment. For example, orders of the processes described herein may be changed and are not limited to the manner described herein. Furthermore, the actions of any flowchart need not be implemented in the order shown; nor do all actions necessarily need to be performed. Also, those actions that are not dependent on other actions may be performed in parallel with the other actions. The scope of the embodiments is in no way limited by these specific examples.Numerous variations, whether explicitly stated in the specification or not, such as differences in structure, dimensions, and material use, are possible. The scope of the embodiments is at least as broad as indicated in the following claims.

[0044] Advantages, other benefits, and solutions to problems have been described above with reference to specific embodiments. However, the advantages, benefits, solutions to problems, and any components that may cause an advantage or solution to occur or become more apparent are not to be construed as a critical, required, or essential feature or component of any or all of the claims. REFERENCES 100 An object detection system for autonomous vehicles, comprising. 102 variety of sensors 104 Sensor fusion module 106 Data processing unit 108 Feature extraction module 110 Object recognition unit 112 decision-makers 114 Calibration module 116 hardware-accelerated processing unit 118 Environmental condition detection unit 120 secure communication interfaces 122 Machine Learning Training Module

Claims

[1] An object detection system for autonomous vehicles, comprising: a plurality of vehicle-mounted sensors, including at least one camera, at least one LiDAR sensor, at least one radar sensor, and at least one ultrasonic sensor, each sensor configured to collect data representative of the environment surrounding the vehicle; a sensor fusion module operatively connected to the plurality of sensors, the sensor fusion module configured to receive and integrate data from the plurality of sensors to generate a comprehensive representation of the vehicle's environment; a data processing unit operatively connected to the sensor fusion module, wherein the data processing unit is configured to preprocess the integrated data by filtering noise, highlighting relevant features, and normalizing sensor outputs to a common reference frame; a feature extraction module operatively connected to the data processing unit, wherein the feature extraction module is configured to identify and extract characteristic features from the preprocessed data corresponding to objects within the vehicle environment; an object detection unit operatively connected to the feature extraction module, wherein the object detection unit comprises a deep neural network model configured to classify the extracted features in real time into predefined object categories such as vehicles, pedestrians, cyclists, traffic signs, and obstacles; and a decision controller operatively connected to the object detection unit, the decision controller configured to analyze the classified objects, determine potential collision risks, and generate appropriate control commands to adjust the navigation path, speed, and other driving parameters of the vehicle to ensure safe and efficient autonomous operation. [2] The autonomous vehicle object detection system of claim 1 further comprises a calibration module operatively connected to the plurality of sensors, the calibration module configured to perform real-time calibration of each sensor to correct for sensor drift, misalignment, and environmental changes, thereby ensuring accurate and consistent data acquisition from the plurality of sensors. [3] An object detection system for autonomous vehicles according to claim 1, further comprising a hardware-accelerated processing unit comprising at least one of a graphics processing unit (GPU) and a tensor processing unit (TPU) operatively connected to the data processing unit and the object detection unit, the hardware-accelerated processing unit configured to process complex computational tasks related to noise filtering, feature extraction, and object classification in parallel, thereby reducing system latency and enabling real-time decision making. [4] The object detection system for autonomous vehicles according to claim 1, further comprising an environmental condition detection unit operatively connected to the data processing unit, the environmental condition detection unit being configured to analyze the data from the plurality of sensors to detect adverse weather conditions such as rain, snow, fog, or low light, and to dynamically adjust the data processing parameters and object detection thresholds to maintain high detection accuracy under varying environmental conditions. [5] An object detection system for autonomous vehicles according to claim 1, wherein the decision controller further comprises a dynamic object tracking module configured to track the movement of classified objects over time using an algorithm based on a Kalman filter or a particle filter, predict future positions and trajectories of moving objects, and adjust the vehicle's control commands to avoid possible collisions based on the predicted trajectories. [6] An object detection system for autonomous vehicles according to claim 1, further comprising a secure communication interface operatively connected to the decision control device, the secure communication interface configured to send and receive data between the autonomous vehicle and a remote server or other autonomous vehicles in the vicinity, thereby enabling cooperative sensing and shared situational awareness to improve overall safety and traffic management in complex driving environments.

Images