Mobility Score Map Fusion for Outdoor Terrain-Aware Driving
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous driving systems using 2D LiDAR struggle to recognize outdoor terrains as obstacles, leading to potential accidents and hardware damage, as they fail to differentiate between drivable and non-drivable areas.
Innovation Solution
A method of generating a score map by merging a vector map with a terrain-based score map, incorporating obstacle, general, and special terrain types, and using this map to control driving routes, avoiding hazardous zones and optimizing speed and torque instructions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a 2D LiDAR-based driving route plan is used, then the mobility can operate in stable indoor environments, but it fails to recognize outdoor terrains accurately, leading to potential accidents
Solution Approach 1:
The patent transitions from 2D LiDAR data to 3D terrain information by introducing elevation data and vertical dimension analysis. The system calculates terrain scores based on height differences, slopes, and vertical obstacles, enabling the mobility to recognize outdoor terrains accurately without adding physical sensors by utilizing the Z-dimension in map data.
Solution Approach 2:
The patent introduces a terrain score map as an intermediary layer between the raw map data and the path planning algorithm. This score map translates complex 3D terrain characteristics into simplified numerical scores that the path planner can process, effectively mediating between detailed terrain information and decision-making systems.
2Device complexity
If the mobility uses basic obstacle detection, then the system complexity remains low, but it misclassifies drivable terrains as obstacles or fails to identify non-drivable areas
Solution Approach 1:
The patent segments the terrain classification into multiple score categories (e.g., safe, cautious, prohibited) based on different terrain characteristics. Each terrain cell is evaluated independently with specific scoring rules for obstacles, slopes, and special terrains, allowing complex classification without requiring a monolithic complex system.
Solution Approach 2:
The patent changes the parameter representation from binary obstacle/non-obstacle to continuous terrain scores. By evaluating multiple parameters (elevation change, slope angle, terrain type) and combining them into composite scores, the system achieves accurate terrain classification while maintaining computational efficiency.
3Productivity
If the mobility plans routes without considering terrain scores, then the path planning is simple and fast, but it may select hazardous routes that damage the vehicle or reduce its lifespan
Solution Approach 1:
The patent performs preliminary terrain score calculation and route evaluation before final path selection. By pre-computing terrain scores for all relevant cells and evaluating multiple candidate routes with penalty scores, the system ensures safe route selection without adding significant computational overhead during real-time operation.
Solution Approach 2:
The patent implements feedback mechanisms where terrain scores and route penalties are continuously updated based on the mobility's position and surrounding conditions. The path planning algorithm uses this feedback to adjust route selection, avoiding hazardous areas while maintaining efficient navigation.
Data Source
AI summary
A method of generating a score map for a mobility can include loading, by a controller of the mobility, a local map including a plurality of cells, scanning, by a surrounding environment scanning unit mounted on the mobility, a surrounding environment of the mobility, recognizing, by the controller, a surrounding terrain of the mobility from information on the surrounding environment of the mobility, generating, by the controller, a first score map corresponding to the terrain, generating, by the controller, a vector map corresponding to the terrain, converting, by the controller, the vector map to a second score map, and generating, by the controller, a final score map based on the first score map and the second score map. A method can further include controlling driving for a mobility. A system of controlling driving of a mobility by using the final score map is also disclosed.