3D Obstacle Trajectory Simulation for Realistic Autonomous Testing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current unmanned simulation scenes do not accurately reflect the obstacle situations encountered by autonomous vehicles during actual driving, as obstacles are manually constructed and do not account for the instability and variability in perception results from sensors, which can impact obstacle avoidance and decision-making algorithms.
Innovation Solution
A method and apparatus that use Gaussian distribution information based on actual perception performance to adjust obstacle motion trajectories and contours in a three-dimensional scene map, simulating the randomness and variability of obstacles to create a more realistic simulation environment, including determining Gaussian distribution parameters and adjusting initial motion trajectories to reflect the actual perception performance of the perception algorithm.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If obstacles are manually constructed in the simulation scene, then the construction process is simple and controllable, but the obstacle information cannot accurately reflect the objective driving scene and perception algorithm performance
Solution Approach 1:
The patent uses sensor data from actual driving scenes to copy real obstacle information into the simulation scene. The perception algorithm processes sensor data to obtain obstacle position, contour, and motion trajectory, which are then directly copied into the three-dimensional scene map, ensuring the simulation accurately reflects real-world conditions without manual construction
Solution Approach 2:
The patent replaces the manual mechanical construction process with an automated information processing system. Instead of manually creating obstacle data, the system automatically extracts obstacle information from sensor data through perception algorithms, substituting manual operations with computational processing to achieve higher accuracy
2Reliability
If actual vehicle testing is performed to verify unmanned vehicle function, then good verification results can be achieved, but the cost and risk are extremely high
Solution Approach 1:
The patent creates a virtual copy of the real driving scene in a three-dimensional simulation environment. By copying obstacle information, road conditions, and environmental data into the simulation scene, the system enables reliable verification of unmanned vehicle functions without exposing actual vehicles to real-world risks and costs
Solution Approach 2:
The patent performs simulation testing before actual vehicle deployment. By preliminarily testing the unmanned vehicle in a virtual environment that replicates real driving conditions, potential issues can be identified and resolved beforehand, preventing costly and dangerous failures during actual operation
3Adaptability or versatility
If perception algorithm performance is considered in obstacle construction, then the simulation can reflect actual driving conditions, but the construction process becomes more complex
Solution Approach 1:
The patent merges the perception algorithm processing steps directly into the obstacle construction process. The perception algorithm processes sensor data to extract obstacle information, which is then immediately used to construct the simulation scene, combining data processing and scene construction into a unified automated workflow that enhances realism without proportionally increasing complexity
Data Source
Figure 1
Figure 2~3
Figure 4
AI summary
The present disclosure discloses a method and an apparatus for simulating an obstacle in an unmanned simulation scene, an electronic device, and a storage medium. The method includes: for obstacle information in a three-dimensional scene map, determining Gaussian distribution information of obstacle position detected by using a perception algorithm to be tested based on actual perception performance of the perception algorithm; adjusting each position of the simulated obstacle in the initial motion trajectory sequence, such that a position deviation between position points in the target motion trajectory sequence and the initial motion trajectory sequence follows a Gaussian distribution; and adding the target motion trajectory sequence to the three-dimensional scene map. In this way, the obstacle information may be accurately constructed in combination with the performance of the perception algorithm in the actual driving scene, and the subsequent simulation test for the unmanned vehicle is facilitated based on the obstacle information.