Test method for self-driving solid-state laser radar sensing algorithm

Abstract

The invention discloses a test method of an automatic driving solid-state laser radar perception algorithm, which belongs to the technical field of automatic driving, and is technically characterized by comprising the following steps of: 1, constructing a multi-scene simulation test environment which comprises different road types, climate conditions, obstacle types and traffic scenes; 2, collecting multi-dimensional data, and carrying out accurate calibration; 3, evaluating the object detection precision, distance estimation error, real-time performance, robustness and data fusion effect of the perception algorithm according to the set performance indexes; 4, carrying out an emergency response test in a dynamic environment, and evaluating the real-time response capability of the sensing algorithm; and step 5, optimizing and adjusting the sensing algorithm according to a test result, and the method has the advantages of ensuring the adaptability and reliability of the system in a complex and dynamic environment by simulating different environmental conditions, road types, weather conditions and dynamic traffic conditions, and promoting the improvement of the safety of an automatic driving technology.

Description

Technical Field

[0001] The present invention relates to the field of autonomous driving technology, and in particular to a testing method for an autonomous driving solid-state laser radar perception algorithm. Background Art

[0002] In the development of autonomous driving technology, lidar, as a key perception sensor, can efficiently obtain three-dimensional spatial information of the environment. Solid-state lidar has become an important part of the autonomous driving system due to its simple structure, low cost, and high reliability. The core task of the solid-state lidar perception algorithm is to extract environmental information from the point cloud data returned by the radar to perform obstacle detection, path planning, autonomous driving decision-making, etc.

[0003] However, existing LiDAR perception algorithm testing methods often focus on the verification of a single scenario or simple environment, and cannot fully evaluate the performance of the algorithm in complex and dynamic environments. This limitation may lead to potential instability or misidentification problems in the perception algorithm in real driving environments, which in turn affects the safety and reliability of the autonomous driving system. Therefore, a new testing method is urgently needed that can comprehensively evaluate the effectiveness and stability of solid-state LiDAR perception algorithms from multiple dimensions, multiple scenarios, and various environmental changes. Summary of the invention

[0004] In view of the deficiencies in the prior art, an embodiment of the present invention aims to provide a testing method for an autonomous driving solid-state lidar perception algorithm to solve the problems in the above-mentioned background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions: A method for testing an autonomous driving solid-state lidar perception algorithm comprises the following steps: Step 1: Build a multi-scenario simulation test environment, including different road types, climate conditions, obstacle types, and traffic scenarios; Step 2: Collect multi-dimensional data and perform precise calibration; Step 3: Evaluate the object detection accuracy, distance estimation error, real-time performance, robustness, and data fusion effect of the perception algorithm according to the set performance indicators; Step 4: Conduct emergency response tests in a dynamic environment to evaluate the real-time response capabilities of the perception algorithm; Step 5: Optimize and adjust the perception algorithm based on the test results.

[0006] As a further solution of the present invention, the testing method also includes a feedback mechanism for evaluating the performance of the perception algorithm after each round of testing and performing iterative optimization based on the feedback results.

[0007] As a further solution of the present invention, the different road types in step 1 include urban roads, rural roads and expressways.

[0008] As a further solution of the present invention, the climate conditions in step 1 include rainy and snowy weather, haze, daytime and nighttime.

[0009] As a further solution of the present invention, the obstacle types in step 1 include vehicles, pedestrians, animals and obstacles.

[0010] As a further solution of the present invention, the traffic scene in step one includes the speed changes of the front vehicle and the side vehicle and the driving conditions at a complex intersection.

[0011] As a further solution of the present invention, the object detection accuracy in step 5 includes obstacle detection rate, false alarm rate and missed detection rate. In summary, compared with the prior art, the embodiments of the present invention have the following beneficial effects: Comprehensiveness: Covers a variety of roads, weather conditions, obstacle types, and dynamic traffic scenarios to test the multi-dimensional performance of the algorithm.

[0012] Efficiency: Through data collection, analysis and feedback mechanisms, potential problems in perception algorithms can be quickly discovered and optimized.

[0013] Accuracy: Accurate test data supports accurate algorithm optimization, improving the detection accuracy and response speed of perception algorithms.

[0014] Robustness: Strengthen the algorithm's adaptability in complex environments and improve the safety of the autonomous driving system.

[0015] In order to more clearly illustrate the structural features and effects of the present invention, the present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a schematic diagram of a test scenario in an embodiment of the invention.

[0017] Figure 2 Schematic diagram of a data acquisition system in an embodiment of the invention.

[0018] Figure 3 It is a test index and performance evaluation diagram in the embodiment of the invention. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0020] The specific implementation of the present invention is described in detail below in conjunction with specific embodiments.

[0021] In one embodiment, a method for testing a solid-state lidar perception algorithm for autonomous driving is described in detail in the following sections. Figure 1 to Figure 3 , including the following steps: Step 1: Build a multi-scenario simulation test environment, including different road types, climate conditions, obstacle types, and traffic scenarios; Step 2: Collect multi-dimensional data and perform precise calibration; Step 3: Evaluate the object detection accuracy, distance estimation error, real-time performance, robustness, and data fusion effect of the perception algorithm according to the set performance indicators; Step 4: Conduct emergency response tests in a dynamic environment to evaluate the real-time response capabilities of the perception algorithm; Step 5: Optimize and adjust the perception algorithm based on the test results.

[0022] For further information, see Figure 1 to Figure 3 ,The testing method also includes a feedback mechanism for evaluating the ,performance of the perception algorithm after each round of testing and performing ,iterative optimization based on the feedback results.

[0023] For further information, see Figure 1 to Figure 3 In step 1, the different road types include urban roads, rural roads and highways.

[0024] For further information, see Figure 1 to Figure 3 The climate conditions in step 1 include rainy and snowy weather, haze, daytime and nighttime.

[0025] For further information, see Figure 1 to Figure 3 , the obstacle types in step 1 include vehicles, pedestrians, animals and obstacles.

[0026] For further information, see Figure 1 to Figure 3 The traffic scene in step 1 includes the speed changes of the front vehicle and the side vehicle and the driving conditions at complex intersections.

[0027] For further information, see Figure 1 to Figure 3 The object detection accuracy in step five includes obstacle detection rate, false alarm rate and missed detection rate.

[0028] In this embodiment, the method first needs to build multiple complex and diverse test environments. The environmental simulation includes but is not limited to the following scenarios: (1) Different road types, such as urban roads, rural roads, and expressways.

[0029] (2) Different climate and lighting conditions such as rain, snow, fog, haze, daytime and nighttime.

[0030] (3) Various obstacle types including vehicles, pedestrians, animals, and obstacles.

[0031] (4) Dynamic traffic environment, including speed changes of the vehicle in front and on the side, and driving conditions at complex intersections.

[0032] We selected typical test sites (such as urban roads, rural roads, and highways) to build test environments under different weather and traffic conditions. For example, when simulating rain and snow, we added artificial rain, fog, and haze simulation equipment to ensure that the algorithm performance under different environmental conditions is fully evaluated.

[0033] 2. Data collection and calibration In the above test environment, data is collected by an actual self-driving vehicle equipped with a solid-state LiDAR. The collected data includes multi-dimensional information such as point cloud data, distance data, and reflection intensity data. At the same time, the LiDAR is precisely calibrated to ensure the accuracy and consistency of the data, so as to provide a reliable data basis for subsequent algorithm evaluation.

[0034] Equipped with a high-precision LiDAR system, it collects point cloud data containing information about obstacles, pedestrians, vehicles, etc. The calibration process eliminates sensor errors and ensures the accuracy of test data.

[0035] 3. Test indicator setting and performance evaluation In order to comprehensively evaluate the performance of the solid-state lidar perception algorithm, this method sets the following key evaluation indicators: (1) Object detection accuracy: including obstacle detection rate, false alarm rate, missed detection rate, etc.

[0036] (2) Distance estimation accuracy: compare the error between the algorithm output and the actual object distance.

[0037] (3) Real-time: Test the processing time of the perception algorithm in different scenarios to ensure that it meets real-time requirements.

[0038] (4) Robustness: Evaluate the stability of the algorithm in complex and dynamic scenarios, including weather changes, traffic emergencies, etc.

[0039] (5) Data fusion effect: Evaluate the data fusion effect of solid-state lidar and other sensors (such as cameras and radars).

[0040] Run the solid-state LiDAR perception algorithm in a test environment, record and analyze the algorithm performance in different scenarios, and conduct quantitative and qualitative evaluations based on the set performance indicators.

[0041] 4. Emergency response testing in dynamic environments Simulate emergency situations that autonomous vehicles may encounter during driving, such as sudden obstacles, sudden braking of vehicles in front, etc. Test the real-time response capability and precise obstacle avoidance capability of the perception algorithm.

[0042] Analyze the algorithm based on the test results, identify key issues and make improvements. For example, optimize the accuracy of object detection, reduce missed detection or false alarms, improve the accuracy of distance estimation, and enhance the effect of data fusion.

[0043] 5. Performance feedback and algorithm optimization During the test, the operating data of the perception algorithm is collected in real time to identify the shortcomings and potential problems of the algorithm and provide optimization feedback. Based on the test results, the perception algorithm is iteratively optimized to improve its performance and stability.

[0044] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A testing method for an autonomous driving solid-state laser radar perception algorithm, characterized in that: The following steps are involved: Step 1: Build a multi-scenario simulation test environment, including different road types, climate conditions, obstacle types, and traffic scenarios; Step 2: Collect multi-dimensional data and perform precise calibration; Step 3: Evaluate the object detection accuracy, distance estimation error, real-time performance, robustness, and data fusion effect of the perception algorithm according to the set performance indicators; Step 4: Conduct emergency response tests in a dynamic environment to evaluate the real-time response capabilities of the perception algorithm; Step 5: Optimize and adjust the perception algorithm based on the test results.

2. The testing method for the autonomous driving solid-state laser radar perception algorithm according to claim 1 is characterized in that: The testing method also includes a feedback mechanism for evaluating the performance of the perception algorithm after each round of testing and performing iterative optimization based on the feedback results.

3. The testing method for the autonomous driving solid-state laser radar perception algorithm according to claim 1, characterized in that: The different road types in step 1 include urban roads, rural roads and expressways.

4. The method for testing the autonomous driving solid-state laser radar perception algorithm according to claim 3, characterized in that: The climate conditions in step 1 include rainy and snowy weather, fog and haze, daytime and nighttime.

5. The method for testing the autonomous driving solid-state laser radar perception algorithm according to claim 4, characterized in that: The obstacle types in step 1 include vehicles, pedestrians, animals and obstacles.

6. The method for testing the autonomous driving solid-state laser radar perception algorithm according to claim 5, characterized in that: The traffic scene in step 1 includes the speed changes of the front vehicle and the side vehicle and the driving conditions at complex intersections.

7. The method for testing the autonomous driving solid-state laser radar perception algorithm according to claim 1, characterized in that: The object detection accuracy in step 5 includes obstacle detection rate, false alarm rate and missed detection rate.

Images