Obstacle Potential Mapping for Vehicle Route Prediction Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing obstacle prediction techniques for self-driving vehicles assume constant acceleration linear motion, leading to inaccurate predictions as actual obstacles do not always follow this motion pattern.
Innovation Solution
An information processing device that acquires obstacle position information, generates obstacle potential maps, predicts moving routes of obstacles, and creates dynamic potential maps considering both stationary and moving obstacles, improving prediction accuracy by accounting for varying motion patterns.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If constant acceleration linear motion assumption is used for obstacle prediction, then calculation simplicity is improved, but prediction accuracy deteriorates
Solution Approach 1:
The patent segments obstacles into two categories: stationary obstacles and moving obstacles. This segmentation allows the system to apply different prediction approaches - simple potential field methods for stationary obstacles and more sophisticated prediction algorithms for moving obstacles, thereby improving overall prediction accuracy without unnecessarily complicating the entire system.
Solution Approach 2:
The patent applies different quality levels of prediction methods to different types of obstacles based on their motion characteristics. For moving obstacles where accuracy is critical, more complex prediction algorithms are used locally, while stationary obstacles continue to use simpler methods. This local differentiation resolves the contradiction by optimizing accuracy where needed while maintaining overall calculation efficiency.
2Productivity
If only moving obstacles are considered in potential map generation, then processing speed is improved, but prediction accuracy for mixed obstacle environments deteriorates
Solution Approach 1:
The patent segments the obstacle potential map generation into two separate components: a stationary obstacle potential map and a moving obstacle potential map. This allows the system to process moving obstacles with full prediction algorithms while handling stationary obstacles through simpler methods, thereby maintaining processing speed while improving accuracy in mixed environments.
Solution Approach 2:
The patent merges the stationary obstacle potential map and moving obstacle potential map to create a comprehensive vehicle potential map. This combination allows the system to benefit from both processing efficiencies (by treating stationary and moving obstacles differently) and complete accuracy (by considering all obstacle types in the final navigation decision).
Data Source
AI summary
An information processing device with improved prediction accuracy of the moving routes of obstacles is obtained. The information processing device according to the present disclosure includes an object position information acquisition unit to acquire obstacle position information showing a position of at least one obstacle existing around a vehicle, an obstacle potential map generation unit to generate an obstacle potential map of risk of a traffic accident for the at least one obstacle based on the obstacle position information, and an obstacle moving route prediction unit to predict a moving route of the at least one obstacle based on the obstacle potential map and generate obstacle moving route prediction information.