Procedural World Generation Using Tertiary Data for AV Simulation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for generating simulated environments for training and testing autonomous vehicles are computationally intensive, time-consuming, and not scalable, requiring manual generation and often taking months or years to produce sufficient data for new geographical locations, limiting the ability to train and validate AI systems before deployment.
Innovation Solution
The use of procedural rendering techniques that leverage sensor data from real environments, supplemented with tertiary data, to generate accurate and scalable simulated worlds, reducing computational resources and time by associating road network data with a road mesh and adding object and surface details procedurally.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If manual generation techniques are used to create simulated environments, then accuracy and detail can be achieved, but the process becomes computationally intensive and time-consuming
Solution Approach 1:
The patent uses real-world sensor data (LIDAR, images, maps) as templates to procedurally generate simulated environments. Instead of manually creating each environment from scratch, the system copies and processes real-world data to automatically generate photorealistic 3D scenes, dramatically improving generation speed while maintaining accuracy through the use of actual sensor measurements
Solution Approach 2:
The system transforms real-world sensor data into simulated environments by changing parameters such as lighting conditions, object positions, and environmental features. By adjusting these parameters procedurally based on real data, the system generates varied accurate environments efficiently without manual intervention for each parameter
2Manufacturing precision
If manual generation techniques are used to create simulated environments, then detailed control can be achieved, but the process becomes time-consuming and not scalable
Solution Approach 1:
The system performs self-service by automatically processing real-world sensor data to generate simulated environments without requiring manual creation. The procedural generation algorithm autonomously extracts features from LIDAR data, images, and maps, and constructs 3D environments, eliminating the time-consuming manual process while preserving detailed accuracy through automated feature extraction
3Manufacturing precision
If existing techniques are used to generate simulated environments, then some level of accuracy can be achieved, but computational resources are excessively consumed
Solution Approach 1:
The system performs preliminary action by collecting and processing real-world sensor data in advance to create reusable 3D models and environmental templates. These pre-processed models can then be efficiently instantiated and modified for multiple simulation scenarios, reducing computational resources required during actual simulation generation while maintaining high accuracy through the use of pre-captured real-world geometry
Data Source
AI summary
Procedural world generation using tertiary data is described. In an example, a computing device can receive road network data associated with the real environment and a road mesh associated with a real environment. The computing device can associate the road network data with the road mesh to generate a simulated environment. Additionally, the computing device can associate supplemental data with the road network data and the road mesh to enhance the simulated environment (e.g., supplementing information otherwise unavailable to the sensor data due to an occlusion). The computing device can output the simulated environment for at least one of testing, validating, or training an algorithm used by an autonomous robotic computing device for at least one of navigating, planning, or decision making.